Machine-learning driven APIs:
Responsible for building the machine-learning driven APIs and Products.
 APIs: General Image Recognition, Sentiment Analysis, Image QA
 Website: https://cognious.com

 □ Implemented Oauth2 authentication for the APIs.
 □ Implemented a distributed task queue using RabbitMQ so that the user doesn't have to wait for the backend tasks to complete.
 □ Implemented memcached to make response faster and efficient.
 □ Implemented access limit, rate limit and built web API access analytics.
 □ Implemented the architecture required for the development and production environment.
 □ Implemented VPN using OpenVPN for secure remote development access.
 □ Built a python Cognious client module for easier API access.
  pypi: https://pypi.python.org/pypi/cognious
  Github: https://github.com/tarunpaul/cognious-client-python

Built a simple android app for Image Recognition.

Personalized News aggregator:
 □ Built a spider to crawl several news sources and aggregate news articles.
 □ Used TF-IDF (Term Frequency–Inverse Document Frequency) algorithm to assign weightage to common words and use it as a feature.
 □ Used NLTK to breakdown the parts of speech and understand the news titles.
 □ Built an ML model to predict the similarity probability of two articles.
 □ Built an external url dispatcher to keep track of the news being read.
 □ Built a simple UI to group and display news.
  https://news.cognious.com/in

URL reach analytics and URL Shortener:
 □ Implemented an Algorithm to generate short url.
 □ Wrote a script to parse Apache logs of the redirect url and gather the data to provide url analytics.
 □ Also implemented a hadoop map reduce code to do the same.

Elevation data analysis:
Made use of the elevation data from SRTM dataset to build the bellow projects:
 □ Build api for elevation data.
 □ Build a map of the low lying areas in Chennai to better analyze the land terrain and the possible effects of flood.
   https://cognious.com/blog/view/chennai-floods-and-terrain-analysis

Home automation:
A home automation project, prototyped using RaspberryPi and Arduino.
 □ Built web api to control all devices from anywhere over internet.
 □ Built a basic android app to control all the devices.
 □ Build a system to turn on lights of a room when motion or human presence is detected.
 □ Built a system to water plants, feed the fishes and control the aquarium light.
 □ Built an ML model to understand voice commands.

Other projects:
 □ Hand gesture detection
 □ AI Game play using OpenAI gym.
 □ A Web Information System for cargo management for Indooverseas Pvt. Ltd

 □ Built several Interfaces (message generation and parser modules)
 □ Re-architectured the Interface message generation and parse engine to generate documentation automatically for each interface which reduces the overall tasks received by the team by 5-7%.
 □ Was a part of a 3 member team, responsible for building Interface Requisition and Information System (IRIS), which tracks and keeps record of all the interfaces we build right from the scope till the end of life cycle of the interface.
 □ Implemented a new version of PIDcheck script to avoid scripts from not starting, by not only clearing all zombie pidfiles but also disguised zombie pidfiles. Using unix file descriptor and created time approach.
 □ In 2 years have delivered 1000+ interface tasks and several hundred peer code check requests.
 □ Received Extra Mile Award for the contribution to the team.