[{"content":"(Originally posted on Medium)\nEvaluating consumer hardware startups\n3 Things hardware startups should focus on As Marc Andreessen famously said; software is eating the world [1]. VCs consider hardware startups riskier than software startups. Hardware startups involve higher costs and have lower margins. The availability of programming tools and internet based services make it easy for software startups to build and validate potential solutions with customers, making them a lower cost, lower risk investment.\nIn spite of the risks, hardware startups have a unique advantage; it’s easier to get people to pay for things they can touch. Connected devices enabling new business models like hardware as a service and subscription models can be highly lucrative. For hardware startups, emphasizing vision and execution strategy is key:\nCreate a solution, not a gadget The focus should be on solving the customer’s problem rather than on new technology. A hardware only product is easy to “rip off,” whereas a solution creates a defensible business. Apple, Tesla, Nest and Fitbit all succeeded because they provided a solution.\nFind your unfair advantage Like any other startup, be clear on the value provided and the core competence. Instead of making everything, focus on building the core and partner with a specialist to form a complete solution.\nIterate faster and cheaper The hardware world needs to embrace the “fail fast” mantra. Hardware tends towards a formal development process, making it slow and tedious to engage with the customer. Multi-project wafer service or 3D printing are excellent means of building cheap prototypes. Creating a ‘minimum viable product’ that can be validated quickly is the key to success.\nReferences [1] http://a16z.com/2016/08/20/why-software-is-eating-the-world\n","permalink":"https://v1vek.com/blog/evaluating-consumer-hardware-startups/","summary":"\u003cp\u003e\u003cem\u003e(Originally posted on \u003ca href=\"https://medium.com/@vzkuma/evaluating-consumer-hardware-startups-ed09246f7c4f\"\u003eMedium\u003c/a\u003e)\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eEvaluating consumer hardware startups\u003c/p\u003e\n\u003ch3 id=\"3-things-hardware-startups-should-focus-on\"\u003e3 Things hardware startups should focus on\u003c/h3\u003e\n\u003cp\u003eAs Marc Andreessen famously said; \u003cem\u003esoftware is eating the world\u003c/em\u003e [1]. VCs consider hardware startups riskier than software startups. Hardware startups involve higher costs and have lower margins. The availability of programming tools and internet based services make it easy for software startups to build and validate potential solutions with customers, making them a lower cost, lower risk investment.\u003c/p\u003e","title":"Evaluating Consumer Hardware Startups"},{"content":"(Originally posted on Medium)\nAlso posted on Quora as an answer to How do I learn deep learning in 2 months?)\n4 Steps for Learning Deep Learning\nA handy list of resources to help you become a deep learning expert Firstly, if you need some basic information or convincing on why Deep Learning is having a significant impact, check out the following video by Andrew Ng\nStep 0: Learn Machine Learning Basics (Optional but recommended)\nStart with Andrew Ng’s Class on machine learning https://www.coursera.org/learn/machine-learning. His course provides an introduction to various Machine Learning algorithms are out there and, more importantly, the general procedures/methods for machine learning, including data preprocessing, hyper-parameter tuning, etc.\nRead the NIPS 2015 Deep Learning Tutorial by Geoff Hinton, Yoshua Bengio, and Yann LeCun for an introduction at a slightly lower level.\nStep 1: Dig into Deep Learning My learning preference is to watch lecture videos and thankfully there are several excellent courses online. Here are few classes I liked\nDeep learning at Oxford 2015 Taught by Nando de Freitas who expertly explains the basics, without overcomplicating it. Start with Lectures 9 if you are already familiar with Neural Networks and want to go deep. He uses Torch framework in his examples. (Videos on Youtube)\nNeural Networks for Machine Learning: Geoffrey Hinton’s class on Coursera. Hinton is an excellent researcher who demonstrated the use of generalized backpropagation algorithm and was crucial to the development of deep learning. I have utmost respect for him, but I found the delivery of this course bit unorganized. Furthermore, coursera messes up with the placement of quizzes.\nNeural Networks Class by Hugo Larochelle: Another excellent course\nYaser Abu-Mostafa’s machine learning course: More theory if you are interested.\nIf you are more into books, here are some excellent resources. Go ahead and check them out, I won’t judge.\nNeural Networks and Deep Learning Book by Michael Nielsen’s: Online book and has several interactive JavaScript elements to play with.\nDeep Learning Book by Ian Goodfellow, Yoshua Bengio and Aaron Courville: Bit denser but never the less a great resource\nStep 10: Pick a focus area and go deeper Identify what you are passionate about and go deeper. The field is vast, so this list is in no way a comprehensive list.\n1. Computer vision Deep learning has transformed this area. Stanford’s CS231N course by Andrej Karpathy’s course is the best course I have come across; CS231n Convolutional Neural Networks for Visual Recognition. It teaches you the basics and up to covnets, as well as helping you to set up GPU instance in AWS. Also, check out Getting Started in Computer Vision by Mostafa S. Ibrahim\n2. Natural Language Processing (NLP) Used for machine translation, question and answering, sentiment analysis. To master this field, an in-depth understanding of both algorithms and the underlying computational properties of natural languages is needed. CS 224N / Ling 284 by Christopher Manning is a great course to get started. CS224d: Deep Learning for Natural Language Processing, another Stanford class by David Socher (founder of MetaMind)is also an excellent course which goes over all the latest Deep learning research related to NLP. For more details see How do I learn Natural Language Processing?\n3. Memory Network (RNN-LSTM) Recent work in combining attention mechanism in LSTM Recurrent Neural networks with external writable memory has meant some interesting work in building systems that can understand, store and retrieve information in a question \u0026amp; answering style. This research area got its start in Dr. Yann Lecun’s Facebook AI lab at NYU. The original paper is on arxiv: Memory Networks. There’re many research variants, datasets, benchmarks, etc that have stemmed from this work, for example, Metamind’s Dynamic Memory Networks for Natural Language Processing\n4. Deep Reinforcement Learning Made famous by AlphaGo, the Go-playing system that recently defeated the strongest Go players in history. David Silver’s (Google Deepmind) Video Lectures on RL and Professor Rich Stutton’s Book is a great place to start. For a gentle introduction to LSTM see Christopher’s post on Understanding LSTM networks \u0026amp; Andrej Karpathy’s The Unreasonable Effectiveness of Recurrent Neural Networks\n5. Generative Models While discriminatory models try to detect, identify and separate things, they end up looking for features which differentiate and do not understand data at a fundamental level. Apart from the short-term applications, generative models provide the potential to automatically learn natural features; categories or dimensions or something else entirely. Out of the three commonly used generative models — Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs) and Autoregressive models (such as PixelRNN) , GAN’s are most popular. To dig deeper read\nOriginal GAN paper by Ian Goodfellow et al.\nThe Laplacian Adversarial Networks (LAPGAN) Paper (LAPGAN) which fixed the stability issue\nThe Deep Convolutional Generative Adversarial Networks (DCGAN) paper and DCGAN Code which can be used to learn a hierarchy of features without any supervision. Also, check out DCGAN used for Image Superresolution\nStep 11: Build Something Doing is key to becoming an expert. Try to build something which interests you and matches your skill level. Here are a few suggestions to get you thinking:\nAs is tradition, start with classifying the MNIST dataset\nTry face detection and classification on ImageNet. If you are up to it, do the ImageNet Challenge 2016.\nDo a Twitter sentiment analysis using RNNs or CNNs\nTeach neural networks to reproduce the artistic style of famous painters (A Neural Algorithm of Artistic Style)\nCompose Music With Recurrent Neural Networks\nPlay ping-pong using Deep Reinforcement Learning\nUse Neural Networks to Rate a selfie\nAutomatically color Black \u0026amp; White pictures using Deep Learning\nFor more inspiration, take a look at CS231n Winter 2016 \u0026amp; Winter 2015 projects. Also keep an eye on the Kaggle and HackerRank competitions for fun stuff and the opportunities to compete and learn.\nAdditional Resources Here are some pointers to help you with continuous learning\nRead some excellent blogs. Both Christopher Olah’s blog \u0026amp; Andrew Karpathy’s Blog do a great job of explaining basic concepts and recent breakthrus\nFollow influencers on Twitter. Here are a few to get started @drfeifei, @ylecun, @karpathy, @AndrewYNg, @Kdnuggets, @OpenAI, @googleresearch. (see: Who to follow on Twitter for machine learning information ? )\nGoogle+ Deep Learning Community Page, by Yann Lecunn, is a good way to keeping in touch with innovations in deep learning as well as communicating with other deep learning professionals and enthusiasts.\nSee ChristosChristofidis/awesome-deep-learning, a curated list of awesome Deep Learning tutorials, projects and communities for more fun\n","permalink":"https://v1vek.com/blog/4-steps-for-learning-deep-learning/","summary":"\u003cp\u003e\u003cem\u003e(Originally posted on \u003ca href=\"https://medium.com/@vzkuma/4-steps-for-learning-deep-learning-86f11fcee54\"\u003eMedium\u003c/a\u003e)\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eAlso posted on Quora as an answer to \u003ca href=\"https://www.quora.com/How-do-I-learn-deep-learning-in-2-months/answer/Vivek-Kumar-893\"\u003eHow do I learn deep learning in 2 months?\u003c/a\u003e)\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e4 Steps for Learning Deep Learning\u003c/p\u003e\n\u003ch3 id=\"a-handy-list-of-resources-to-help-you-become-a-deep-learning-expert\"\u003eA handy list of resources to help you become a deep learning expert\u003c/h3\u003e\n\u003cp\u003eFirstly, if you need some basic information or convincing on why Deep Learning is having a significant impact, check out the following video by Andrew Ng\u003c/p\u003e\n\u003cp\u003e\u003ca href=\"https://www.youtube.com/watch?v=n1ViNeWhC24\"\u003e\u003cimg alt=\"Video\" loading=\"lazy\" src=\"http://img.youtube.com/vi/n1ViNeWhC24/hqdefault.jpg\"\u003e\u003c/a\u003e\u003c/p\u003e\n\u003ch2 id=\"step-0-learn-machine-learning-basics\"\u003e\u003cem\u003e\u003cstrong\u003eStep 0: Learn Machine Learning Basics\u003c/strong\u003e\u003c/em\u003e\u003c/h2\u003e\n\u003cp\u003e\u003cem\u003e(Optional but recommended)\u003c/em\u003e\u003c/p\u003e","title":"4 Steps for Learning Deep Learning"},{"content":"(Originally posted on Medium)\nSocial network is not the most interesting part. The keynotes and highlights of the conferences were other parts — Messenger, Instagram, WhatsApp, and Oculus. Facebook, like Apple is doing a great job of disrupting itself, mainly through acquisitions. Yes even Messenger was an acquisition\nFacebook’s 10-year road map\n“Facebook’s 10-year roadmap is basically lasers, bots and VR” — Napier Lopez The Bots are Coming Facebook released Chatbot API’s making it easy to create Bots without native Apps. Additionally the Wit.ai’s Bot Engine enables natural language communication in both text and speech. By feeding the engine new information, new tasks, and new data, it will understand more situations and will become smarter over time. Facebook already uses state-of-the-art AI in moments and in curation of News Feed.\nAR and VR Throughout F8 there was a big focus on making VR (and in future ,AR) the next big social network innovation for both for gaming and long distance social interactions. The clunky form factor is limiting adoption right now but in 10 years headsets are predicted to evolve into “Ray-Ban” like devices.\nFacebooks also debuted their open source 360 VR camera and system software. The camera will cost upwards of $25,000 to build, includes 17 different synchronized capture devices and can record two hours of 360-degree video at up to 60 frames per second, as much as 8k per eye.\nIn the future, you’ll be able to snap your fingers and pull out a photo and make it as big as you want. … A lot of the things we think about as physical objects — like a TV — will actually just be one dollar apps in the app store . — Mark Zuckerberg Connected World There was also quite a focus on connecting the world. It was once again highlighted that almost 60% of the world\u0026rsquo;s population still doesn’t have internet access. The major barriers to worldwide Internet connectivity cited were availability (affecting one billion people) , affordability (one billion), and awareness (two billion). The solutions touted ranged from solar-powered drones, lasers and satellite-based service. The controversial free Basics program for developing countries was also touted once again.\nOther highlights Other noteworthy announcements made at F8 2016\nSharing Button — which developers will be able to customize, will allow users to share quotes they find interesting from around the web directly on Facebook.\nSave to Facebook — save information from other apps and websites.\nOpening Livestream API — enabling devices other than a standard smartphone to stream to Facebook Live. During the keynote, Zuck live streamed video from a drone as a demonstration\nOther Coverages of F8 Videos - Facebook for Developers Edit descriptiondevelopers.facebook.com The Facebook Federation *Reading over the coverage of F8 this week, one thing is clear: Facebook the social network isn’t very interesting…*medium.com Messenger Platform at F8 *Launching the Messenger Platform (Beta) Every month, over 900 million people around the world communicate with friends…*newsroom.fb.com 6 Big Takeaways From Mark Zuckerberg’s Facebook F8 Keynote *Facebook’s mostly-annual developers conference, F8, kicked off in San Francisco today and will run through tomorrow…*www.inverse.com\n","permalink":"https://v1vek.com/blog/insights-from-f8-developer-conference/","summary":"\u003cp\u003e\u003cem\u003e(Originally posted on \u003ca href=\"https://medium.com/@vzkuma/insights-from-f8-developer-conference-87116c2957c9\"\u003eMedium\u003c/a\u003e)\u003c/em\u003e\u003c/p\u003e\n\u003cblockquote\u003e\n\u003ch1 id=\"social-network-is-not-the-most-interesting-part\"\u003e\u003cstrong\u003eSocial network is not the most interesting part.\u003c/strong\u003e\u003c/h1\u003e\n\u003c/blockquote\u003e\n\u003cp\u003eThe keynotes and highlights of the conferences were other parts — Messenger, Instagram, WhatsApp, and Oculus. Facebook, like Apple is doing a great job of disrupting itself, mainly through acquisitions. Yes even Messenger was an \u003ca href=\"http://techcrunch.com/2011/03/01/facebook-beluga/\"\u003eacquisition\u003c/a\u003e\u003c/p\u003e\n\u003cp\u003e\u003cimg alt=\"Facebook’s 10-year road map\" loading=\"lazy\" src=\"https://cdn-images-1.medium.com/max/2000/1*kOz2fWyrOd_BwpLTe-vo1A.png\"\u003e\u003cem\u003eFacebook’s 10-year road map\u003c/em\u003e\u003c/p\u003e\n\u003cblockquote\u003e\n\u003ch1 id=\"facebooks-10-year-roadmap-is-basically-lasers-bots-and-vr--napier-lopez\"\u003e“Facebook’s 10-year roadmap is basically lasers, bots and VR” — \u003ca href=\"undefined\"\u003eNapier Lopez\u003c/a\u003e\u003c/h1\u003e\n\u003c/blockquote\u003e\n\u003ch2 id=\"the-bots-are-coming\"\u003e\u003cstrong\u003eThe Bots are Coming\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003eFacebook released Chatbot API’s making it easy to create Bots without native Apps. Additionally the Wit.ai’s Bot Engine enables natural language communication in both text and speech. By feeding the engine new information, new tasks, and new data, it will understand more situations and will become smarter over time. Facebook already uses state-of-the-art AI in moments and in curation of News Feed.\u003c/p\u003e","title":"Insights from F8 Developer Conference"},{"content":"(Originally posted on Medium)\nBased on photo from https://www.flickr.com/photos/jdlasica/12338690004\nBuilding a Business When There Are No Easy Answers Ben Horowitz is the cofounder of Andreessen Horowitz (A16Z) and one of Silicon Valley’s most respected and experienced entrepreneurs. In ‘The Hard Thing about Hard Things’ he shares his experience of being a founder-CEO and the hard decisions he has had to make — offering advice on managing tough problems as a leader, which business schools do not cover. While most management books focus on doing things correctly, Ben offers insights into what you must do after you have screwed up.\nQuoting Ben,\nThere’s no recipe for building a high-tech company; there’s no recipe for leading a group of people out of trouble; there’s no recipe for making a series of hit songs; there’s no recipe for playing NFL quarterback; there’s no recipe for running for president; and there’s no recipe for motivating teams when your business has turned to crap. That’s the hard thing about hard things — there is no formula for dealing with them\nHorowitz was exposed to a diverse perspective from a very early age; parents indoctrinated in the Communist philosophy, grandparents were card-carrying Communists and at Berkeley High School football team, he was the only player to excel academically. Looking at the world through different perspectives helped him learn how to separate facts from perceptions. This ability proved incredibly useful when he became an entrepreneur, especially in dire circumstances when looking at alternative narratives and radically different perspectives kept him informed.\nHorowitz founded a company called LoudCloud in 1999, based on the emerging concept of a computing cloud\nIn spite of the challenges, Horowitz firmly believed in “Treating your employees fairly and always telling them the truth.” He understood that figuring out the right product is the innovator’s job, not the customer’s and after the first line of code is written, decisions are no longer objective in the technology business.\nChapter 4: When Things Fall Apart When the odds of LoudCloud’s survival were low, Ben’s friend and LoudCloud board member, Bill Campbell advised him to prepare for bankruptcy, but he never made such a contingency plan. He believed that CEOs should not play the odds. When you are building a company, you must believe there is an answer, and you cannot pay attention to the odds of finding it. You just have to find it. It matters not whether your chances are nine in ten or one in a thousand; the task is the same.\nThere is no secret to being a successful CEO he states; however, the ability to focus and make the best move when there seems to be no good moves is crucial. When things get unbearably difficult, and the struggle begins, there are a few things you can try.\nFirst, do not put it all on only your shoulders. You should share every burden that you can with the maximum number of brains.\nSecond, CEO’s Should Tell It Like It Is and ensure informal information and ideas are flowing freely in your company, instead of adopting the old management standard: “Don’t bring me a problem without bringing me a solution.” What if the employee cannot solve an important problem?\nThirdly — and this is a tough one — you will have to lay people off at some point. Hence, get your head right by focusing on the future of the company rather than the past. Do not delay. If word leaks out before you execute the decision, a whole new set of issues will arise. Be clear about why you need to lay people off. If it is because the company failed to hit its plan, then admit that failure. Then be visible and present, to boost the confidence of those who are still on board.\nSo, what’s the secret to being a “successful” CEO? Sadly, none, Ben says, other than making the best move when there are no good moves.\nHororwitz puts forward the idea that there is a significant difference between a Peacetime CEO vs. a Wartime CEO, which sparks great interest. Let me quote directly from the book:\nPeacetime CEO knows that proper protocol leads to winning. Wartime CEO violates protocol in order to win. Peacetime CEO focuses on the big picture and empowers her people to make detailed decisions. Wartime CEO cares about a speck of dust on a gnat’s ass if it interferes with the prime directive. Peacetime CEO builds scalable, high-volume recruiting machines. Wartime CEO does that, but also builds HR organizations that can execute layoffs. Peacetime CEO spends time defining the culture. Wartime CEO lets the war define the culture. Peacetime CEO always has a contingency plan. Wartime CEO knows that sometimes you gotta roll a hard six. Peacetime CEO knows what to do with a big advantage. Wartime CEO is paranoid. Peacetime CEO strives not to use profanity. Wartime CEO sometimes uses profanity purposefully. Peacetime CEO thinks of the competition as other ships in a big ocean that may never engage. Wartime CEO thinks the competition is sneaking into her house and trying to kidnap her children. Peacetime CEO aims to expand the market. Wartime CEO aims to win the market. Peacetime CEO strives to tolerate deviations from the plan when coupled with effort and creativity. Wartime CEO is completely intolerant. Peacetime CEO does not raise her voice. Wartime CEO rarely speaks in a normal tone. Peacetime CEO works to minimize conflict. Wartime CEO heightens the contradictions. Peacetime CEO strives for broad-based buy-in. Wartime CEO neither indulges consensus building nor tolerates disagreements. Peacetime CEO sets big, hairy, audacious goals. Wartime CEO is too busy fighting the enemy to read management books written by consultants who have never managed a fruit stand. Peacetime CEO trains her employees to ensure satisfaction and career development. Wartime CEO trains her employees so they don’t get their asses shot off in the battle. Peacetime CEO has rules like “We’re going to exit all businesses where we’re not number one or two.” Wartime CEO often has no businesses that are number one or two and therefore does not have the luxury of following that rule.\nChapter 5: Take Care of the People, The Products, and the Profits — in That Order According to Horowitz, it is crucial to create a good work environment. It’s important that the people who spend most of their waking life at work have a good environment to exist in while there. Being a good company doesn’t matter when things go well, but it can be the difference between life and death when things go wrong.When things go well, there are many reasons to stay at a company —\nYour career grows as the company grows, attractive jobs naturally open up\nYou’ll be impressing your friends and family. Your friends and family will think you a genius for choosing to work at the “it” company\nYour résumé gets stronger by working at a blue-chip company in its heyday.\nOh, and you are getting rich.\nHowever, when things go poorly, all those reasons evaporate and become reasons to leave. In fact, the only thing that keeps an employee at a company when things go horribly wrong — other than needing a job — is that she likes her job, despite the hardships\nSo on that note the book advises when expanding your team, consider:\nHire for strength rather than lack of weakness.\nHave clear expectations of who you are hiring with a realization that there is something seriously wrong with every employee in your company (including you). Nobody is perfect.\nInvolve multiple people in brainstorming but make the final decision solo. Consensus-based decisions tend to sway the process away from strength and towards weakness.\nEspecially when hiring executives, it is very important to remember that being a big company executive is very different from being a small company executive. Big company executives are driven by interruptions, while startup executives know that nothing happens unless they make it happen. A big company executive has an incredible number of incoming demands on his time, whereas in a small company without a massive push, the company will stay at rest.\nChapter 6: Concerning the Going Concern I think Horowitz is quite correct when he states in this chapter that most workplaces are far from good. As organizations grow larger, important work can go unnoticed. The hardest workers can get passed over in favor of the best office politicians, and bureaucratic processes can choke out the creativity and true productivity.\nAs a company grows, Horowitz says, it is important to minimize politics within your organization. One of the biggest issues behind politics is when people advance their own careers or their personal agendas instead of merit or contribution.\nHere are the four takeaways for minimizing politics in the company.\nFirstly, hire people with the right kind of ambition; otherwise, your company would turn into the political equivalent of the U.S Senate. The right kind is ambition for the company’s success; with the executive’s own success coming on only as a by-product.\nSecondly, maintain strict policies and processes on organizational design, performance evaluations, promotions, and compensation\nThirdly, you should promote experienced employees by measuring results against objectives, management skills, innovation, and their ability to work well with others.\nFourthly, ensure one-on-one meetings between employees and managers. These are an excellent platform for employees to discuss their as yet unheard brilliant ideas, pressing issues, and chronic frustrations.\nChapter 7: How to Lead Even When You Don’t Know Where You Are Every CEO, Horowitz writes, of a startup should focus on what needs to be right rather than worrying about what is wrong. One of the most difficult challenges is keeping your mind in check. You need to be able to move aggressively and decisively without acting insane. To calm your nerves, find someone you can talk to who understands what you’re going through. Put your ideas, challenges, and fears on paper for a better focus on where you are going and not what you are trying to avoid.\nThere is a fine Line Between Fear and Courage — People who watch you judge you on what you do, not how you feel.\nChapter 8: First Rule of Entrepreneurship: There Are No Rules Reiterating a theme that comes up again and again in the book as a whole this chapter is devoted to the idea that there are no rules in business. Things may seem to be going well, but they can change in an instant. You need accountability and creativity to succeed in business. Accountability is the key for effort, promises and results.\nThe hard things will always be hard; courage and grit are the keys to success.\nChapter 9: The End of the Beginning Continuing his personal story Horowitz recalls that after selling Opsware, he went to work for Hewlett-Packard, but he still knew he wanted to do something else. He decided to set up a firm designed to help technical founders run their companies. Technical founders are the best people to run technology companies. All long-lasting technologies thrived when led by their innovator ; Intel, Amazon, Apple, Google and Facebook had been run by their founders and in some cases still are to this day\nWith this new company Horowitz wanted to imapart to founders that things are hard because there are no easy answers or recipes. They are hard because your emotions are at odds with your logic. They are hard because you don’t know the answer and you cannot ask for help without showing weakness. While his firm cannot give a founder CEO all the skills he or she needs, it can provide mentorship based on his experience and learnings.\nSummary While most start-up stories may read like a romance novel, this tome is more like watching the film “Fight Club”. The writing has a biting, sarcastic, blunt humor and uses hip-hop lyrics as intros to chapters. It’s fun and entertaining to read but most of all it’s an amazing insider look into the world of tech and how to succeed in this unique industry.\nIf you are leading a team of any size there is much more to be learned from “The Hard Thing About Hard Things’ and so I would strongly recommend getting and reading a copy for yourself.\n","permalink":"https://v1vek.com/blog/the-hard-thing-about-hard-things-summary/","summary":"\u003cp\u003e\u003cem\u003e(Originally posted on \u003ca href=\"https://medium.com/@vzkuma/the-hard-thing-about-hard-things-summary-eab9b51c420a\"\u003eMedium\u003c/a\u003e)\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eBased on photo from \u003ca href=\"https://www.flickr.com/photos/jdlasica/12338690004\"\u003ehttps://www.flickr.com/photos/jdlasica/12338690004\u003c/a\u003e\u003c/p\u003e\n\u003ch3 id=\"building-a-business-when-there-are-no-easy-answers\"\u003eBuilding a Business When There Are No Easy Answers\u003c/h3\u003e\n\u003cp\u003e\u003ca href=\"http://www.bhorowitz.com\"\u003eBen Horowitz\u003c/a\u003e is the cofounder of \u003ca href=\"http://a16z.com\"\u003eAndreessen Horowitz\u003c/a\u003e (A16Z) and one of Silicon Valley’s most respected and experienced entrepreneurs. In ‘\u003ca href=\"http://www.amazon.com/The-Hard-Thing-About-Things/dp/0062273205\"\u003eThe Hard Thing about Hard Things\u003c/a\u003e’ he shares his experience of being a founder-CEO and the hard decisions he has had to make — offering advice on managing tough problems as a leader, which business schools do not cover. While most management books focus on doing things correctly, Ben offers insights into what you must do after you have screwed up.\u003c/p\u003e","title":"The Hard Thing About Hard Things — Summary"},{"content":"(Originally posted on Medium)\nTop 5 Videos from Extract Conference 2015\nExtract conference, 2015 was a day jam-packed with data stories meant to entertain, educate and inspire. The stories were a combination of data analytics and growth hacking told by the people who know it the best. They are all interesting and here are my favorite top 5 videos for Data Stories Worth Sharing.\nDeep Learning — Andrew Ng Andrew Ng, Chief Scientist at Baidu, talks about recent developments in Deep Learning and the work being done by Baidu. This session explores what deep learning is all about and why it matters in 20 minutes — liquid gold. If you had to see just one presentation from the conference this would be it. The slides are posted here: http://www.slideshare.net/ExtractConf\n2. Kaggle — Anthony Goldbloom Anthony Goldbloom, Founder \u0026amp; CEO of Kaggle, gives an overview of Kaggle. Kaggle is a platform for machine learning competitions on which companies and researchers post their data and statisticians and data miners from all over the world compete to produce the best models. Since the early days of Kaggle, several companies have used the competition to find elegant solutions for complex real-life problems. He also discussed the recent shift in the use of different algorithms over time; recently xgboost+ensembles are winning most of the recent competitions. Most of the code is now posted at Kaggle Scripts, which can run code and access datasets from the browser without downloading anything, not even the data. Check out some cool examples here https://www.kaggle.com/scripts\n3. The Beautiful Science of Data Visualization — Jeff Pettiross Jeff Pettiross, Head of UX at Tableau, discusses the history and science of Data Visualizations. I really liked his venture into the history of Visualization with Minard’s cartographic depiction of numerical data on a map of Napoleon’s disastrous losses suffered during the Russian campaign of 1812. He also discussed recent scientific advances in Data Visualization — what works and how to use it.\n4. Google Search Ranking Factors — 2015, Rand Fishkin Rand Fisher, Founder of SEO powerhouse Moz, discusses how Googles search ranking operates and what factors really matter. In an effort to make Google’s search algorithm more transparent, he discusses how he used correlation data to extract what matters.\n5. Collecting Startup Funding Data — Sarah Catanzaro Sarah Catanzaro, Head of Data Science at Mattermark, describes their journey using Machine Learning to Collect Startup Funding Data. She goes over the transition to ML from manual data collection and offers an overview of data scrubbing, human support and validation and their learnings.\nBonus: drizzy.ai — Matt Ellsworth Yes I know, I was trying to pick just thetop 5 but Matt’s presentation on Recreating a Rapper From Scratch was too good to pass up. Matt is a Distribution Partner at 500 Startups and is funny as hell. He narrates his journey of creating an intelligence entity for Comedy Hack Day 15, which tweets and responds like rapper Drake.\nThese videos were quite enlightening and if you are interested in seeing rest of the videos here’s the complete playlist\nExtract Conference, San Francisco 2015 Videos ","permalink":"https://v1vek.com/blog/top-5-videos-from-extract-conference-2015/","summary":"\u003cp\u003e\u003cem\u003e(Originally posted on \u003ca href=\"https://medium.com/@vzkuma/top-5-videos-from-extract-conference-2015-b3184bbb281e\"\u003eMedium\u003c/a\u003e)\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eTop 5 Videos from Extract Conference 2015\u003c/p\u003e\n\u003cp\u003e\u003ca href=\"http://extractconf.com/\"\u003eExtract conference\u003c/a\u003e, 2015 was a day jam-packed with data stories meant to entertain, educate and inspire. The stories were a combination of data analytics and growth hacking told by the people who know it the best. They are all interesting and here are my favorite top 5 videos for Data Stories Worth Sharing.\u003c/p\u003e\n\u003col\u003e\n\u003cli\u003eDeep Learning — Andrew Ng\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003e\u003ca href=\"https://www.youtube.com/watch?v=videoseries\"\u003e\u003cimg alt=\"Video\" loading=\"lazy\" src=\"http://img.youtube.com/vi/videoseries/hqdefault.jpg\"\u003e\u003c/a\u003e\u003c/p\u003e\n\u003cp\u003eAndrew Ng, Chief Scientist at Baidu, talks about recent developments in Deep Learning and the work being done by Baidu. This session explores what deep learning is all about and why it matters in 20 minutes — liquid gold. If you had to see just one presentation from the conference this would be it. The slides are posted here: \u003ca href=\"http://www.slideshare.net/ExtractConf\"\u003ehttp://www.slideshare.net/ExtractConf\u003c/a\u003e\u003c/p\u003e","title":"Top 5 Videos from Extract Conference 2015"},{"content":"(Originally posted on Medium)\nDesign Thinking and the Art of Innovation\nI have always believed that everybody can increase their creativity if they put their mind to it, and the recent course at Stanford University on Design Thinking and the Art of Innovation (XDES400–019) provided the essential tools to do just that. I was inspired to take this course after watching David Kelley’s TED talk on How to build your creative confidence.\nThe class was led by Bill Burnett, the executive director of d.school who has worked in both start-ups and Fortune 100 companies, including seven years at Apple. Here are my key takeaways from the course.\nBrainstorming We have all tried brainstorming at some point, but the approach in d.school was scientific; unlike traditional approaches.\nPractice Makes Perfect: Creative thinking in general and brainstorming, in particular, are skills that need to be practiced. Burnett mentioned Gladwell’s’ 10,000 Hour Rule to achieve expertise, with continual practicing using right approaches. He also referenced an article in the New Yorker by Jonah Lehrer called ‘Groupthink: The Brainstorming Myth’. The article argued that individual ideation produces more creative ideas than brainstorming, which Burnett mentioned was expected of novices in the skills.\nWarmup: Like you have to warm-up before hitting, the gym, the brain also needs flexing before it gets in the Zone. We experienced Several improvisation activities /games to warm up as a group. The point of these exercises is not to get good results, but to make you break away from your other tasks, discussions, and everything else that is occupying your mind. This also helps in ensuring that no idea is shut down and the imagination runs amok.\nYes and…: During brainstorming try to push boundaries, encourage wild ideas, Ensure no idea is shut down, and there is no judgment. Let the imagination run amok. Push the limits of impossible and simulate “possible” ideas that may be rather out of the box. The focus should be on quantity and not quality. Build upon other people’s ideas by using a Yes and … technique.\nSelection Criteria: Here, Burnett said, is where most groups fall short. He suggested using colored dots to votes on ideas most likely to succeed, most likely to delight and ideas that would create the greatest breakthrough.\nPrototyping for Empathy Several times during the course, participants were asked to make something physical to test with their users. These prototypes were made in less than 10 minutes using everyday materials like paper, aluminum foil, and pipe cleaners.\nAn amazing array of designs emerged and a strong creative energy lit up the room. When the prototypes were shared to validate insights, there was a great sense of genuinely being listened. Several participants came up with prototypes showing true potential, and I would not be surprised to see few of them on Kickstarter soon.\nSo the biggest takeaways were: understand, observe, visualize and iterate. The trick here is, that the big deal is the iteration. Rather than planning incessantly you quickly come up with something smart, you show it to users, and then you do it again and again.\nTour of IDEO We believe great innovators and leaders need to be great design thinkers.\nDavid Kelley The first day of training ended with a tour of IDEO and a Creative Confidence talk by David Kelley, founder of IDEO and Stanford’s d.school. Walking around IDEO brought home the importance and impact of physical spaces and how rapid prototyping directs the innovation process. Kelley reiterated the importance of human-centric design, quick iterations, and building on other’s ideas. Loved the signed copy of his book.\nSummary Above all, when taking this course I loved the fact I got to meet and interact with diverse, talented people from various backgrounds, some starting their venture and some planning on impacting innovation practices in their organization. Unarguably the best course I have taken in recent history.\nCheck out his TED talk on how to build creative confidence, the one that inspired me to take this course in the first place.\n","permalink":"https://v1vek.com/blog/design-thinking-and-the-art-of-innovation/","summary":"\u003cp\u003e\u003cem\u003e(Originally posted on \u003ca href=\"https://medium.com/@vzkuma/design-thinking-and-the-art-of-innovation-dcbf89da433d\"\u003eMedium\u003c/a\u003e)\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eDesign Thinking and the Art of Innovation\u003c/p\u003e\n\u003cp\u003eI have always believed that everybody can increase their creativity if they put their mind to it, and the recent course at Stanford University on \u003ca href=\"http://scpd.stanford.edu/search/publicCourseSearchDetails.do?method=load\u0026amp;courseId=6085755\"\u003eDesign Thinking and the Art of Innovation (XDES400–019)\u003c/a\u003e provided the essential tools to do just that. I was inspired to take this course after watching \u003ca href=\"https://www.ted.com/speakers/david_kelley\"\u003eDavid Kelley\u003c/a\u003e’s TED talk on \u003ca href=\"https://www.ted.com/talks/david_kelley_how_to_build_your_creative_confidence\"\u003eHow to build your creative confidence\u003c/a\u003e.\u003c/p\u003e\n\u003cp\u003eThe class was led by \u003ca href=\"https://www.billburnett.net/\"\u003eBill Burnett\u003c/a\u003e, the executive director of d.school who has worked in both start-ups and Fortune 100 companies, including seven years at Apple. Here are my key takeaways from the course.\u003c/p\u003e","title":"Design Thinking and the Art of Innovation"},{"content":"Investment Thesis I focus on startups with speculative technologies and a founding team with passion and capability to disrupt the status quo. Innovative companies that cross the technical uncertainty chasm and create disruptive technology end up controlling their destinies. That is what I hope to help founders build.\nTrends I am excited about:\nDemocratization of Technology: The transition to a bottom-up economy and democratization of technology is creating opportunities across industries. As software continues to eat the world, a trend that continues with mobile access and AI presents opportunities for innovative startups to disrupt across verticals and lead for years to come.\nAttention Economy: Time is the most valuable skill we have in the attention economy, and I am incredibly positive about the potential of Artificial Intelligence (AI) to simplify our lives.\nPortfolio Company Description Link Voicery Synthesizing realistic human voices using deep neural networks voicery.com CitrusLabs A new way of patient recruitment citruslabs.io Squad Screen sharing with friends squadapp.io Proof Website personalization made easy useproof.com Auris AI platform for business decisions aurius.io Tangent Personalizing beauty with breakthrough AI tangent.ai Profiles: Wellfound · Crunchbase\n","permalink":"https://v1vek.com/angel/","summary":"Angel investments by Vivek Kumar.","title":"Angel Investing"},{"content":"See Google Scholar for a complete listing of papers and patents.\nTalks Audio AI - Challenges, Breakthroughs \u0026amp; Applications - PyTorch DevCon 2019 Artificial Intelligence in Audio – Applications, Advancements And Trends - Soho - 477 Broadway, New York, 2019 [info] Future of Audio - IEEE MIPR, Santa Clara, 2019 BISH Bash Meetup: Audio AI - Hosted by Dolby Laboratories, San Francisco, 2019 Challenges of Doing Deep Learning with Audio - REWORK Applied AI Summit, San Francisco, 2018 When Deep Learning Takes Over Signal Processing - Machine Learning Summit, San Francisco, 2018 [slides] Learning Deep Learning - Deep Learning Meetup, 2017 Papers PhaseCoder: Microphone Geometry-Agnostic Spatial Audio Understanding for Multimodal LLMs, Artem Dementyev, Wazeer Zulfikar, Sinan Hersek, Pascal Getreuer, Anurag Kumar, Vivek Kumar, ICML, 2026 [pdf] [bibtex]\nSpeechCompass: Enhancing Mobile Captioning with Diarization and Directional Guidance via Multi-Microphone Localization, Artem Dementyev, Wazeer Zulfikar, Sinan Hersek, Vivek Kumar, arXiv preprint, 2025 [pdf]\nSPAE: Semantic Pyramid AutoEncoder for Multimodal Generation with Frozen LLMs, Lijun Yu, Yong Cheng, Zhiruo Wang, Vivek Kumar, Wolfgang Macherey, Yanping Huang, David A. Ross, Irfan Essa, Yonatan Bisk, Ming-Hsuan Yang, Kevin P. Murphy, Alexander G. Hauptmann, Lu Jiang, NeurIPS, 2023 [pdf]\nVoice conversion with conditional SampleRNN, Interspeech, 2018 [pdf]\nTransform-domain decorrelation in Dolby Digital Plus, IEEE ICASSP, 2014 [pdf]\nPseudo-Reliable Code Development, Embedded Design, 2000 [pdf]\nPatents Systems and methods for adapting human speaker embeddings in speech synthesis, US Patent 11,929,058, 2024 Speech style transfer, US Patent 11,538,455, 2022 Audio capture for aerial devices, US Patent 10,979,613, 2021 Low bit rate parametric encoding and transport of haptic-tactile signals, US Patent application WO2017024001A1, 2017 Adaptive quantization, US Patent application WO2017132366A1, 2017 Time-varying filters for generating decorrelation signals, US Patent application 2014126684, 2014 Signal decorrelation in an audio processing system, US Patent application US201443877, 2014 Bit error concealment for audio coding system, US Patent US8301440B2, 2012 Real time monitoring \u0026amp; control for audio device, US Patent US7778829B2, 2010 Sampling rate mismatch solution, US Patent US7778373B2, 2010 Bit error management methods for wireless audio communication channel, US Patent US8578247B2, 2013 Method and apparatus for sharing a bluetooth module with two computing devices, US Patent 7,263,331, 2007 ","permalink":"https://v1vek.com/research/","summary":"Talks, publications, and patents by Vivek Kumar.","title":"Talks and Papers"}]