Best of 2016

Here’s my top 5 list of Books, Movies, and Sporting moments from 2016. The top five are in no particular order. I did end up reading about 50 books this year and did not go through the trouble of listing out everything. Most categories have a ‘missed’ section where I call out misses. I saw all the top sporting moments live on TV. Lastly, I might do a top 5 Music list in the next couple of days.


  1. Shoe Dog
  2. The Boys in the Boat
  3. Something incredibly wonderful happens: Frank Oppenheimer
  4. The Guernsey literary and potato peel pie society
  5. The three body problem
  1. The Black Prince of Florence
  2. Surely you’re joking Mr.Feynman
  3. The Wright Brothers
  4. Levels of the Game

Movies (English/International)

  1. Zootopia
  2. Sully
  3. Deadpool
  4. Hacksaw Ridge
  5. Elle (French)
  1. Doctor Strange
  2. Kubo and the two strings
  3. Batman v Superman
  4. Equity
  5. Carol
  6. Rogue One
  7. Jungle Book
  8. No Country for Old men
  9. Groundhog day
  10. London is falling
  11. Taxi (Iranian)
  12. Kung Fu Panda 3
  13. The Conjuring 2
  14. Pele: Birth of a legend
  15. 13 hours: Secret soldiers of Benghazi
  16. Toni Erdmann (German)
  1. Elvis & Nixon (missed)
  2. Moonlight (missed)
  3. Arrival (missed)

Movies (Hindi)

  1. Dangal
  2. Kapoor and Sons
  3. Ae Dil Hai Mushkil
  4. Neerja
  5. Dear Zindagi
  1. MS Dhoni
  2. Te3n
  3. Baar baar dekho
  4. Airlift
  5. Ki & Ka
  6. Brahman Naman
  7. Udta Punjab
  1. Pink (missed)

Movies (Tamil)

  1. Jil Jung Juk
  2. Kabali
  3. Irudhi Sutru
  4. Aandavan Kattalai
  5. Chennai 28 2nd innings
  1. Kodi
  2. Kadhalum Kadandhu Pogum
  3. Sethupathi
  4. 24
  5. Devi
  6. Iru Mugan
  7. Accham Yenbadhu Madamaiyada
  8. Aviyal
  9. Amma Kanakku
  10. Appa
  11. Iraivi


  1. The grand tour
  2. Game of thrones
  3. Stranger things
  4. The night of
  5. Mozart in the jungle
  1. Naruto Shippuden
  2. Narcos
  3. Catastrophe
  4. The Crown
  5. Master of None
  6. The Night Manager
  1. Halt & Catch Fire (missed)


  1. Cavs winning the NBA
  2. Cubs winning the World Series
  3. Ronaldo winning the Euro
  4. Serena tying with Steffi @ 22 slams
  5. Reed v Rory @ the Ryder
  1. Bolt 3/3
  2. Kobe’s last game
  3. Phelps’ 200 m IM Gold
  1. Leicester winning the Premier League (missed)

Sales v Product

Is it tough to ‘sell’ products today? Companies that have traditionally poured 80% of their resources into Sales are looking to allocate the same percentage to Product instead.

Are today’s prospects exponentially more cynical about Sales reps? Has this shift in resource allocation contributed to consumer surplus? Consumers perhaps have a subconscious understanding of the existence of this surplus. The same absolute dollar investment in Product is bound to be more efficient (can scale better and marginal cost is zero) than an investment in hiring a Sales team (not to mention the challenges of developing a comp plan that works).

Not sure if consumers realize this but the price they pay for a ton of products/services today is significantly lower thanks to productivity gains driven by technology. The GDP metric does not capture any of this.

So then, should the public not worry about stagnant incomes?

I think not yet. Startups have to solve for ‘safety’ in Maslow’s hierarchy. We need some superstar entrepreneurs to solve for education and healthcare in partnership with the government. I am personally more optimistic about healthcare. Machine learning might hold the key to cure the Baumol’s cost disease plaguing both these areas.

A critical bottleneck to unlock the power of ML will be the availability of data to train models. But if ML is able to create the ultimate impact in these two industries what of the huge displacement of jobs (including white collar)?

Looks like this is not a Sales v Product post at all.


Tracking sleep #Fitbit

I cannot bring myself to wear a Fitbit while sleeping. It doesn’t seem natural. Perhaps it’s just me or is it the case that a majority of Fitbit owners don’t track sleep?

I am curious to see if this can be solved w/o the need to develop a different tech solution. But first, is this a problem that’s worth solving for Fitbit? Let’s say they solve it and by the flip of a switch a deluge of sleep data gets collected. What’s in it for Fitbit? I think they have already done their job, really well I must add.

What’s the end game for quantified self companies? Will we see data monetization strategies being deployed soon? Imagine a world where marketing will be tailored based on your place in the health curve. Maybe the mattress you sleep on sucks and Sleep Number might send something relevant your way. Is that a legitimate channel to reach consumers?

If tracking sleep is tough I can only imagine the difficulty with tracking food. We definitely need a tech based solution for that. Eating the right stuff has a huge impact on your health and unfortunately we don’t yet have a great way to force that behavior is my opinion.



dfxMachina #dealmemo #hardware #factory

What would the widget factory of the future look like? Today, consumer electronics manufacturing plants and assembly lines are highly manual where labor cost is a big line item. Hardware companies are facing increasing competition and are hence forced to squeeze their contractors/sub-contractors to reduce costs. But where did that leave the factories in China? Many Chinese factories resorted to buying robots that could do the job. But, that didn’t pan out, electronics manufacturing is a much more intricate business compared to automobiles and other industrial scale equipment. Further, the end consumer’ tolerance for faulty product is zilch, for good reason. But soon, all that’s going to change.

While the contractors have been busy throwing money at robots the world of software has been chipping away at computer vision. The driving force is a field called machine learning. Machine learning is when a computer learns by example instead of by strict rules that have been programmed. Specifically, there are algorithms called neural networks, deep neural networks, or deep learning that have been making huge advancements in the field. More recently, many of the techniques that have been gaining traction are called “deep learning.” Deep learning is machine learning, but instead of looking for simple patterns, it is able to look for patterns-of-patterns. For example, if a machine sees a Fitbit, maybe it can start to put together the LED display or holes in the strap from edges and curves. And then finally, it can put together all the patterns that make up the Fitbit pieces into a whole device.

Why do I bring this up? Deep learning plus machine aided manufacturing is the key to lower skilled labor costs in the contractor ecosystem. It will also help reduce cycle-time, improve accuracy as well as repeatability, and finally eliminate process/re-work waste. Investment in industrial robots has happened but the software investment didn’t keep up, until now. Expect to see a huge uptick in data gathering and ML specific investments.

Enter dfxMachina as the deus ex machina of hardware startups. The stealthish company (backed by Root Ventures) is building an automated factory operating system. dfxMachina is founded by two former Apple mechanical engineers – including the lead on Apple Watch. Instead of just making products thinner/cheaper/better, they want to fundamentally improve how all products are built. dfxMachina’s technology will level the playing field for Davids fighting Goliaths in the hardware development space. dfxMachina is the hero hardware startups need and deserve. 

Hopscotch #DealMemo #coding #edtech

I was chatting with a computer science PhD friend a few months ago and she said something that struck me. She said that anyone can be a great programmer, you just need the 3Cs, 1) computer, 2) command-line, and 3) composure. And I thought this helps explain why many students of programming are confused by the sustained failure of their attempts to talk to a machine. I am guilty of being impatient myself and have asked my developer friends on multiple occasions “Why can’t coding become a drag-and-drop type of an exercise w/ some instant gratification thrown in?” and “Why not code on the supercomputer all of us carry in our pockets all the time?”

Enter Hopscotch. They seem to have answered my questions. I don’t see Hopscotch a la a Codecademy or Khan Academy, because as good as those companies are their coding UX is commoditized. I see potential in Hopscotch to become a highly engaged community of coding enthusiasts, something that will be very hard for competitors to replicate.

My two favorite games growing up were Dangerous Dave and Prince of Persia. I was obsessed. That would have probably been a great time to get me hooked onto developing similar games and teaching core coding concepts in the process. Hopscotch unfortunately wasn’t around when I was 10 but today young people use their mobile block-based app to easily program games and software. Each week, their users publish over 100K projects that are played over 2MM times. Seen through the lens of my friends’ comments, it is a great user engagement plan to target a young audience in the 21st century – mobile only, drag & drop interface, instant artisanal indie games, all to democratize the coding experience.

Why should parents and schools be enthusiastic about using Hopscotch to teach kids programming? When I was growing up, we were told to reduce our time playing games on computers and focus instead on analog activities. But we should have been maybe told the opposite. Games are a great way to prepare kids for the future. A future where communicating with machines will be a reality. Creating games is as hard as it gets when it comes to computer programming and is a great way to activate both left and right brain.

Hopscotch’s CEO and co-founder is Jocelyn Leavitt who has been working on this full time for ~3 years. The CTO and co-founder is Samantha John who has been full-time for ~5 years. Looks like a good combination of tech and business chops. I am also psyched that Jocelyn shares my alma mater (Go Big Green!). It appears that both co-founders share a passion for teaching and that that passion has culminated in Hopscotch.

The K-12 education market is huge in the US (~$650B I read somewhere). It will be interesting to see how much of that is addressable and then further capturable for Hopscotch. The obvious routes seems to be partnership with forward thinking traditional schools, tutors, and new concepts like AltSchool. Another great avenue will be parents (the oldest millennial cohort) who work at the largest tech companies and appreciate the product and its impact intuitively. From the outside, I’d love to see Hopscotch launch an Android app quick on the heels of the recent iPhone app launch. The revenue model seems to be based on in-app purchases currently but might need experimenting with subscription and/or ad based, for example. This is where partnerships will certainly help.

The near term challenge will be to figure out a way to not alienate the innovators who typically demand powerful features and balance this request against the need to cater the product towards new users. Eliminating the urge to be many things to many audiences will get the company to profitability. Prioritization will be everything in the goal to profitability, customer feedback coupled with company values will inform these priorities.

I am super excited about Hopscotch and will be watching them do great things in the computer programming education space.

Pefin #DealMemo #fintech #finAI

Today I’m excited to write about a fintech startup Pefin, Pefin’s mission is to provide unbiased and affordable financial planning and advice to anyone seeking it. Pefin will use Artificial Intelligence to make decisions with your money, just like a human advisor would.

The CEO and founder of Pefin is Ramya Joseph. Ramya concurrently completed her MS in Machine Learning (AI) and a MS in Financial Engineering from Columbia University. She appears to have a deep understanding of portfolio construction and trading cost optimization. Her technical co-founder is Joe Abraham and has expertise in software architecture, development and project execution of large-scale systems.

Pefin’s technology tracks more factors than the software used by human financial advisors — market rates, tax codes, inflation, area-specific property taxes, and more. On average, there are over 2M data points analyzed for each user every time a new plan is added or market conditions change.

Financial advisory startups have historically limited themselves to servicing customers cheaper than incumbents. But “cheaper” isn’t interesting, it’s not something that creates disruptive change. I think there will be lots of changes in financial technology, but only small businesses will come out of it, no really big businesses. This brings us to the need for a full-stack fintech startup.

As we move around in this heavily connected world, all the data about where we go, what we eat and where we shop, act as bread crumbs we leave behind, and eventually, as we fill in more details through check-ins and updates such as tweets, a pattern of our spending emerges. While harbingers of privacy might find it alarming, there is another way to look at all of this. The tapping and analysis of this data is the key to a smarter way to manage finances than a spreadsheet or piece of paper.

There is a ton of direct competition in the fintech advisor space with companies like wallet.AI, Betterment, Ellevest, FutureAdvisor and Wealthfront. My instinct tells me that a true full-stack startup that integrates tax returns, remittances, investments, budgeting, planning, lending etc will own the fintech category. Technology and competition will only drive margins lower so a full suite of solutions will be a great hedge to possess but tough to execute.

I do not have data regarding the progress that Pefin has made since launch early this year and am yet to try out the product (they appear to have >$5M from undisclosed investors). But there is something alluring about stealth companies, the team has been working on this for > 5 years now. I am eager to see what Pefin can achieve over the coming years. They appear to have a strong team and are focused on product differentiation, both of which is the need of the hour in the fintech space.

Avametric #DealMemo

Yet another post-Series A startup that looks really promising. I am really excited about what this company can do to disrupt online apparel shopping.


Avametric is building a technology to help online shoppers see how a garment will fit on a 3-D digital image of themselves, enabling them to virtually try on clothes online, and switch between different sizes and colors to pick the perfect look. It’s VFX meets retail e-commerce. For now.

Product/market fit

The product couldn’t launch at a better time. I hear from a bunch of millennial consumers that they hate buying clothes online purely because of the hassles around returning items that don’t fit (even if the returns are paid for). For the retailers, the costs of these returns are an unnecessary expense item. The market will pull this product in. There also exists an orthogonal opportunity to extend into bespoke and tailored clothing in the future.

It speaks volumes that the company already have over 80 partnership deals with brands. Once Avametrics launches with these brands, brace yourselves for an online sales avalanche.

Market size

The U.S. apparel market is the largest in the world, comprising about 28 percent of the global total and has a market value of about 331 billion USD. I found this chart on U.S. apparel and accessories retail e-commerce revenue from 2013 to 2019 (in billion USD).

Apparel ecom

This is what I believe to be the total addressable market for Avametric. But there are obvious additive opportunities including partnering with niche fashion brands that are not yet online and helping them embrace e-commerce.


I am a huge fan of founder-CEOs so a little disappointed to see that’s not the case here. That said, stellar team on paper. Eclectic selection of VFX gurus, CS talent, and fashion merchandising vets. The founder is still with the company and is the CTO. The former head of Google Glass, Ivy Ross is advising the team as an independent board member. But the biggest draw for me is that Keith Rabois is backing the team. He is unbelievably amazing.


I believe that Avametric at scale will possess non-obvious network effects due to the immense data points they collect. At the same time it is a challenge to build a data business, something which requires tremendous focus and long-term thinking. But as the tech market matures, products like Avametric will become the infrastructure for the new economy.


Macy’s and Nordstrom are both hurting from the Amazon onslaught. Avametric will accelerate this trend. If the company continues to strike partnerships with tech savvy brands and e-commerce juggernauts like Amazon, the future is bleak for traditional fashion retail leaders. On the other hand this is an opportunity for Macy’s/Nordstrom to plug the leaky bucket before it’s too late. Either way Avametric is in pole position to capitalize and potentially become a platform on top of the retail stack and perhaps go direct-to-brands eventually to obtain a larger slice of the pie.


This is a b2b business and if the product or service is sold on a subscription basis, then the amount and timing of expected churn will be important to track . I’d have to guess what the price elasticity of the service is. But in a subscription offering,


These are exciting times as we move into the next phase of innovation in online retail and I think Avametric is going to be a huge disruptor in apparel. Such paradigm shifting products are rare (no historical precedent) and a focus on execution (business development/partnerships) will set them up for success.

Learn more about Avametric.