University of Florida, Gainesville, FL

Master of Science, Computer Science (expected Fall, 2018)

Software Developer Intern, backend, RealTruck.com (Florida, USA)

www.realtruck.com/

Technologies: Python, Node JS, Tensorflow, Google Compute Engine and Machine Learning Engine, serverless, AWS Lambda,SQL

• Developed a product recommendation system by building machine learning models using Tensorflow
• Queried Amazon Redshift Database for relevant user-data, such as purchase history, location, etc.
• Deployed the said models into production on Google Cloud Platform
• Built REST APIs and clients in Node Js to extend and access model’s predictions to send out automated mass promotional emails to existing users

Machine Learning Research Intern, Indian Institute of Technology (Delhi, India)

www.iitd.ac.in/

• Designed and implemented machine learning algorithms to automatically segment out tumor regions from Diffusion-weighted MRI scan images. Dice Coefficient - 68%
• Outcome - developed software is being used by doctors at India’s best Medical Institution, AIIMS, Delhi

Winner - BestHack, Swamphacks, 2018 [ Android, TensorFlow, PyTorch, Node JS, VPS, Deep Learning]

swamphacks-iv.devpost.com/submissions

Open Doors: An Accessibility-First application that helps blind people navigate busy lobbies using their phone’s camera

• Designed and implemented Deep Learning models using Tensorflow and PyTorch SDKs
• Developed Node Js service to support real-time video predictions

Finalist, VandyHacks, 2017 [ Android, Microsoft Cognitive Services, Microsoft Azure Face API, Youtube V3 API, Web Scraping]

Musical Faces: Mobile application that recommends music based on user’s facial expressions

• Consumed Microsoft Cognitive Services API and Azure Face API to detect user’s facial expressions
• Built a network utility to scrape dynamically updating genre based music lists from the web
• Embedded Youtube player and consumed youtube API for in-app streaming of recommendations 3

Database Engine Implementation, [ C++, pthreads, Gnu Debugger, lex, yacc, Valgrind ]

Built a Database system from ground up that supports a subset of SQL functionality

• Implemented multithreaded External Two-Pass Multiway Merge Sort algorithm to support record Sorting
• Implemented paging to reduce number of disk-accesses and developed memory optimal relational operations
• Designed and incorporated query optimisation based on relational data
• Used pthreads for concurrency, e.g. handling the producer-consumer infrastructure, etc.

Distributed System - Gossiper , [ Elixir, ErLang, Gossip Protocol, Actor Model]

• Built a Distributed, scalable, fault tolerant information sharing system based on the Actor model for concurrency, implementing the Gossip and Push-Sum protocols
• Tested the gossiper on various network topologies, such as linear, 2-D grid, Sphere/Torus, Full network of more than 10,000 nodes

Sentiment Analysis, [ Java, Neural Networks, Stanford Core NLP, Natural Language Processing]

Sentie: Implemented Dense Neural Networks from scratch along with Natural Language Processing techniques such as n-gram models and parts-of-speech tagging to predict sentiments of paragraphs of text. Dataset - http://sentiwordnet.isti.cnr.it/