Few of my projects.

CNN Collaborative Filtering Cuda Deployment Django Image Processing Kinect Machine Learning Microservices MongoDB Naive bayes classifier Neural Networks Object recogintion Python RabbitMQ Sentiment Analysis Swift Tensorflow Torch Xcode iOS macOS

Hand detection and tracking

Jan 2020-Present
Working with Machine common sense project funded by DARPA
• Deployed an ML model based on Faster-RCNN to detect hands and further working on fine tuning the model.
• Implemented tracking algorithm to track detected hands in a video frame
Python Torch

Financial named entity recognition

Jun 2018-Jun 2019
Work done @ Societe Generale
• Developed and implemented Bi-LSTM based NER to extract finance context-based data from scanned documents. Worked on state of the art Flair library for improving net performance
• Integrated and deployed several microservices using RabbitMQ
• Managed overall operations and deployment of ML-based microservices and databases over a cluster of servers for high scalability and availability.
Python Tensorflow RabbitMQ Microservices Django MongoDB

Google Summer of Code 2016

April-August 2016
Aim of this project was to integrate of Unity Game scene with the existing pyKinect and emulate a moving skeleton based of the movements tracked by the Kinect sensors.
Python Kinect

Google Summer of Code 2015

April-August 2015
The current Summer of code project is to deploy planning and scheduling for rovers using Europa-PSO which is developed by NASA. After the plans are made and solved by the solver, it passes the instructions to rover for execution.

Google Summer of Code 2014

April-August 2014
Developed a voice interface which detects voice commands through a microphone that is connected to a Raspberry pi. For voice recognition, Pocket sphinx was used which is developed by Carnegie Mellon University . The parsed voice commands are then sent as commands to the rover which performs the respective action.Went to Italy for final testing and integration.

Object Recoginition using CNN.

Sept-Oct 2016
This project uses CNN to detect certain shapes/markers in the images. More info on github page.
Machine Learning Cuda CNN Object recogintion

Signature Verification using ANN.

Oct 2015
This project was a research work done on signature analysis which utilizes Java for feature extraction from images of signatures and uses a neural network coded in Python from ground up for classification of signatures.

The important role in this project was to make a neural network that would classify the features extracted from the signatures. These features were used as inputs to the neural network and the output neurons were according to the number of signatures that were taken into consideration.
Python Image Processing Machine learning Neural Networks

Smart Job recommendation

July-December 2015
This project was about a job recommendation system based on collaborative filtering technique for job portals.The system is designed to suggest the jobs to the user depending upon his profile and by calculating a similarity index using Euclidian distance of two skill sets and then ranking them according to their naïve Bayes algorithm.
Collaborative Filtering Naive bayes classifier

iOS App on VTU syllabus

Oct 2015
• This simple application had a local database of syllabus of all the subjects of VTU.
• First of its kind on the App Store.
• 2500+ downloads on app store.
Swift iOS Xcode macOS Deployment


Oct 2014
• Extract business comments/tweets and analyse them.
• Based on the analysis, rate the news as positive, negative or neutral (Sentiment analysis).
Sentiment Analysis