Python is my goto language for most tasks. 
I have used Python to write machine learning algorithms, do complicated analysis of images/text/audio data, manage the backend for an app (db management and microservices), write many RESTFUL APIs, interconnect many APIs (Google analytics, Spotify, Slack, Duolingo,Google Maps, etc) and write web scrapers.

It is easy to read, has an amazingly large collection of very well written/maintained packages and gets me from 0 to 100 in a matter of hours rather than days or months. 

I have published data analysis studies (analysing millions of data points) which have gone viral and been featured in The Guardian, CNN, NPR, Mtv and other print/digital publications. Many of these have hit the front page of Reddit as well.
I believe that a simple and insightful analysis of data supported by a visualization and a crisp explanation can bring forth many hidden and even surprising aspects of data. 

I love to solve real world problems with machine learning. I have used support vector machines, decision trees and neural networks working as a Researcher with Yamaha Music and Musixmatch. These projects were built in Python using scikit-learn and related libraries. Some examples - automating phonetic segmentation of singing voice in audio files, prediction of song mood using audio/lyrics, detecting singing voice in songs, etc. 

I am constantly trying out and reading about new algorithms and models that advance the state of the art in machine learning. 


I have used Javascript and nodejs to do front end development for web pages which have seen 1000+ users in realtime without any hiccups. Coupling various libraries using npm and controlling the dev/build processes using webpack has allowed me to make clean/responsive webpages quickly. Most of these webpages were for visualizing complicated datasets. 
I also turn to Javascript on the backend sometimes for tasks which need async capabilities which can't be easily achieved with Python.

I worked at Musixmatch, the most downloaded lyrics app and the largest lyrics provider worldwide, as a Research Engineer. 
My main projects were based on using state of the art algorithms on this lyrics database which spanned many languages. Some of these projects were mood prediction using lyrics and audio, recommendation system based on similarity analysis of lyrics, filtering 'bad' lyrics, detecting instrumental songs using audio, etc.
Many of these projects are integrated in the Musixmatch apps or on the backend to deliver a unique and clean experience to the users.
I love the challenge of writing algorithms to extract insights from structured and unstructured text.


I use flask whenever I need to serve any kind of data through a REST API. Combined with Elastic Beanstalk, I can get a scalable API running in a matter of minutes.

I have been using various AWS services in production for over 3 years. All the servers I use are EC2 based. The ease of setup/use and performance of these servers continues to amaze me. 

Most of my Python based APIs are hosted using Elastic Beanstalk. This allows me to easily scale the APIs based on incoming requests and takes a lot of sys-admin work off my plate.

Overall, learning to interconnect and use various AWS tools has been a game changer for me as a developer and also for the various companies where I have worked. 

Over the last two years convolutional neural networks and deep learning have immensely advanced the state of the art of image recognition, language processing and various other research fields. In the last year or so we have seen various products - google photos/translate, Prisma, MSQRD, etc - putting these advances in the hands of millions of users. 
In my latest project, I have been fortunate to learn and implement a CNN based system for categorizing images and an image to image search. This system uses transfer learning over the famous googlenet model. It is in the beta testing phase and will soon be released as a part of a shopping app.

I love the extreme flexibility and speed of MongoDB (nosql dbs in general). This is my preferred database for most tasks when I need to get up and running quickly.

Having said that I have also optimised/managed MongoDBs in production settings and learn new features almost each day.

Personal details

Varun J.

Summary

Technical skills