Hi! I'm Debarghya Das . People call me Deedy .
I read. I write. Here's my blog.

I've been a Software Engineer at Google in New York City
since January 2016 working on knowledge graph, query understanding and personalization for Search.

I spent a year at Facebook working on locations and places.

I graduated with a Bachelors and a Masters in
Computer Science from Cornell University in 2014.

I maintain a list of the stuff I enjoy reading, watching, or learning from called where the knowledge is and occassionally write on my blog, Writes.

I've worked at a few places too.

Jan '16 - Present

I'm helping organize the world's information in the New York City office, on the Search team working with machine learning and infrastructure for knowledge graph, query understanding and personalization for Search.

Jan '15 - Jan '16

I helped make the world a more open and connected place at the New York City office, with tons of data and machine learning. I was on the Places Search Ranking team, and hacked on locations, geo, and search.

Cornell Robot Learning Lab
Jan '14 - Dec '14

I did research here, broadly in Robotics and Machine Learning, under Professor Ashutosh Saxena. I recently published a paper as well. Most of the team now works out of Stanford on Robobrain.

Cornell University
Aug '11 - Dec '14

I went to college here, in Ithaca. I learnt, did research and helped teach. It's cold and beautiful. You walk a lot. I graduated in three and half years with a Bachelors and a Masters.

Jun - Sep '14

I interned in Mountain View on the Future Learning Exprience team as a full-stack engineer and a KPCB Fellow. I shipped Yoda, and had a tremendously fun summer being mentored by the incredible Jon Wong.

May - Aug '13

I interned at the YouTube Captions team in Mountain View under Loretta Guarino Reid. I shipped the Captions editor and creator to production and met some really smart people.

Jan - May '13

I contributed to this open-source project as a part of a Facebook initiative, met the legendary Evan Priestley and amongst harder things, created the meme generator.

La Martiniere for Boys
May '04 - May '11

I went to school here, in Kolkata, for 8 years. It was founded in 1836, so it's pretty old. It turned 175 when I graduated. I was taught Math by David Royan.

Some stuff I did gained some popularity.

In June '13, I published Hacking into the Indian Education System about some code I wrote to expose glaring anomalies in one of India's premier education boards. It amassed a lot of media coverage.

In August '13, I wrote a Quora answer about the perks of interning at Google, following the release of the movie The Internship (2013) about the same, which featured on Business Insider.

In April '14, another Quora answer about the impact of competitive programming on careers featured in Forbes.

Yet another Quora answer from July '13 about college advice students are not likely to ever hear was featured on Business Insider.

In April '14, I open-sourced a XeLaTeX resume template that went viral and has over 1200 stars and 350 forks on the Github repo after trending on Reddit and HackerNews. It features on many popular online TeXing platforms and archives.

It is one of the top 10 most popular Latex templates on Overleaf, with over 85,000 views.
It initially peaked at number 5 on HackerNews.
The Open-Source community has made a Resume Generator based on it that peaked at number 2 on HN and a version in Jade.

In August '14, I spoke at TEDx Bangalore to an audience of 1600 about some major issues plaguing Indian education and how online education can be and is being an effective solution. The video is available here.

I've several talks at some of India's best engineering colleges.

about software, technology, my projects and my past experiences.

I was on Al Jazeera live on February '17 to talk about United States H-1B visa (temporary worker visa) policy along with Bruce Morrison, former Congressman and the primary drafter of the H-1B bill of 1990, and several others. The telecast can be found here.

For speaker invitations, I can be reached by email or on facebook.

Research is fun.

I've been fortunate to have been in involved with research with Prof Ashutosh Saxena in Robotics and Machine Learning, Sam Tilsen in Linguistics, and Kavita Bala in Graphics and got two research papers published.

PlanIt: A Crowdsourcing Approach for Learning to Plan Paths from Large Scale Preference Feedback
Ashesh Jain, Debarghya Das, Ashutosh Saxena

Tech Report, Dec 2014
IntInternational Conference on Robotics and Automation (ICRA), 2015

Real-time articulatory biofeedback with electromagnetic articulography
Sam Tilsen, Debarghya Das, Bruce McKee

Linguistics Vanguard, Dec 2014

I making cool shit.

Which Delhi University college will I get into?
Jan '17

Delhi University is one of India's biggest universities with offering over 50 subjects amongst more than 50 colleges. Admissions to DU are based off of complicated cut-off calculations, innumerable quotas, and one needs to look at multiple cut-off list pdfs being served through websites stuck in the 90s. I scraped and structured the data from DU admissions and provide a clean, user-friendly form so you can check which colleges and for what subjects you're eligible to be admitted in.

The Grad School Admission Statistics We Never Had
Nov '15

GradCafe has 372 thousand graduate school admissions results. I scraped 93% of that data, including 80 thousand GRE scores and 75 thousand undergraduate GPAs. I also extensively deduplicated the user provided university names to over 98% cleanliness, and compiled a hyperclean Computer Science specific set of 28 thousand results. Let's dive into these never-before-seen statistics on graduate school admissions! Retweeted and shared by many professors.

Looking back at 9 years of Hacker News
Nov '15

Hacker News, the Y Combinator tech forum, has exploded from its humble beginnings in 2006 to one of the primary sources of news in the tech community. Looking through the data, we can see how HN grew, the things people talked about, the most shared domains, the most influential contributors and more. Received over 100k hits.

The Crazy Christian Coder
Spring '14

Inspired by King James Programming, I decided to write my own generic Hidden Markov Model that generates funny one liners by doing a random walk between two (or more) works of text. Here is a sample from the mix between King James Bible and Artificial Intelligence: A Modern Approach. And here's one with the latter and the English translation of Mahabharata. They're hilarious.

A computer is mostly limes with chords of the terrified, herd of deer.

Fall '14

Torrage is a super secure peer-to-peer cloud storage platform written in C, where your password never leaves your machine and your files never leave your machine in one piece or unencrypted. It's based on the BitTorrent protocol. Read the paper here. I'm going to open source the code soon.

The Indian Voter
Spring '14

Worked with tGELF to aggregate and analyze over one and a half million tweets related to the 2014 Indian General Elections. We visualized interesting trends with time, location, and sentiment.

Fantasy Cricket Trading Algorithm
Spring '14

I developed a player trading model for the Fantasy Indian Premier League, the world's biggest cricket league. It was the first such model to compete in the actual fantasy league. It used Binary Integer Programming, computationally outperforming previous research in the field by 33x and achieved 99.54%ile out of over 400,000 participants. Read the paper here.

Spring '14

Marauder allows players to create and join maps that continuously but efficiently log location data from your phone and displays live locations of all players on the map with very low latency, much like the Marauder's Map. We used Tornado, WebSocket, and Redis amongst other technologies, and built a web app and iOS app.

Latex Resume Template
Spring '14

Finally ported my obsolete Word resume into Latex and open-sourced it! There aren't many half-decent resume templates out there. I also compiled a few of my favorite Latex templates as well.
The template has over 700 stars and is the 4th most popular English resume creating system on Github.

Twin Trading Models
Spring '14

I developed two trading models - a volatility sensitive Markowitz model for small investors, as well as a Support Vector Machine model limited to trading only on SPDR ETFs, managing to outperform baselines and in the first case, half the risk. Read the paper here.
This paper was tweeted by Carl Carrie, former Global Head of Algorithmic Products at JP Morgan and Head of Commodity Technology at Barclays.

Cornell Robot Learning Lab
Spring '14

Worked with Ashesh Jain and Ashutosh Saxena to build PlanIt, a Machine Learning system that helps robots plan paths in human environments from crowd-sourced user preference data. Read the paper here.

Connect4 Artificial Intelligence
Fall '13

Used Play! to make a Java-backed deterministic Minimax Tree based AI, with optimizations enabling a 13 move lookahead. Deployed on Heroku.

Fall '13

I worked in a team of 4 to create a compiler from scratch in Java for Cubex. We used ANTLR, and implemented some optimizations. Ross Tate taught us and he's awesome.

Cornell Phonetics Lab
Spring '12 - Spring '13

Wrote code which used the WAVE sensor system to eventually create the world's first accurate tongue controlled game. A paper is in progress.

Summer '13

A Chrome extension and mobile website that allows you to use any mobile device as a remote for YouTube.

Summer '13

Monitors your vital signs with HumanAPI. Performs sentiment analysis on your Facebook conversations to find who cheers you up and connects you with them when it detects stress.

Summer '13

Uses governmental data about farm land to use 7 different factors to algorithmically rank land in the USA by quality depending on what the farmer wants to rear or grow using ArcGis.

Winter '12

Minimalistic instant search tool that uses the RottenTomatoes to quickly find synopses, User ratings and Audience ratings.

Violet Satellite Project Team
Fall '11 - Spring '12

Wrote a GUI, an orbit analysis tool, and did some code factoring on a simulation for a nanosatellite that will actually launch into space.

Winter '12

Captor is a demo of a YouTube tool which allows quick, intuitive video captioning. This was the prototype for the actual YouTube Captions Creator.

Other Coursework-based Projects

I've also created a fully rendered, realistic and interactive 3D Game with a random terrain generator, a Ray Tracer renderer, a multithreaded SMTP Server, my own malloc, a fully functional filesystem with a multithreaded garbage collector, a MapReduce in OCaml, a Pokémon playing bot, a MIPS Processor, and Breakout.

Other Personal Projects

I've also created a simple algorithmic trading simulator, tools to scrape the Cornell University student database, a Typeracer hack, a Cornell instant course search tool called Instudy, classic games like Pong, Tetris and Minesweeper, and a 3D Graphing utility.

And a little bit about me...

I enjoy critically reasoning and debating controversial topics. Topics that particularly interest me are India, education, politics, technology, and economics.

My Myers-Briggs Type Indicator(MBTI) is INTP [Introverted iNtuitive Thinking Perceiving]. The favorite people I supposedly share my personality type with are Albert Einstein, Larry Page and Sergey Brin. You too should take the test here.

I write on Quora whenever I feel like I have something new to add, usually about India, life, code and education. I really want to start my own blog soon but haven't quite gotten around to it. and I finally did.

I enjoy hiking, cricket, eating and reading about tech. I'm compiling a list of my favorite books, videos, articles and news sources on the internet in a list called where the knowledge is.

I love Hindi music and have a large Spotify playlist going, cleverly marketed as the Best Hindi Playlist on Spotify. I also like Keane and Michael Jackson and have recently gotten into music from different countries. When working, I listen to ambient storm sounds on Noisli.

I once lived off coffee, beer, and 5-hour energy shots, often in that order.
Now it's mostly honey green tea and cheap wine.

And here's my resumé...