Academics

Awards/ Achievements

• Non student Travel Grant – EMNLP, 2018
• ISCA Student Travel Grant – Interspeech, 2018
• Undergraduate Research Award – IIT Bombay, 2018
• Institute Academic Award for exceptional academic performance – IIT Bombay, 2015
• AP (Advanced Performer) grade for exceptional performance in Programming Langauges and Engineering Drawing
• KVPY fellowrship – Government of India 2014
• All India Rank 154 in JEE Advanced 2014
• All India Rank 93 in JEE Mains 2014

Relevant Courses

• Machine Learning :
  • Web Search: Web Search and Mining, Organization of Web Information
  • NLP/Speech: Automatic Speech Recognition, Introduction to Natural Language Processing
  • Image Processing: Medical Image Processing, Digital Image Processing
  • Fundamentals: Optimization, Foundations of Machine Learning, Artificial Intelligence, Data Analysis and Interpretation, Introduction to Probability Theory, Statistical Inference, Regression Analysis, Linear Algebra, Numerical Analysis
• Computer Science : Implementation of programming languages, Operating Systems, Computer Networks, Database and Information Systems, Computer Architecture, Automata theory and logic, Discrete Structures, Software Systems Lab, Data Structures and Algorithms, Design and Analysis of Algorithms, Logic for Computer Science, Digital Logic Design, Abstraction and Paradigms in programming

Academic Projects

• From Image Classification to Audio Classification | Guide: Prof. Preethi Jyothi, Autumn 2017
  • Used embeddings from trained image classification architectures (viz. VGG net) to classify the audio
  • Implemented multiscale input representation using dilated convolutions to transfer learned representations from the image classification task to an unrelated task of voice focused audio classification.
• FCNs for Semantic Segmentation | Guide: Prof. Suyash Awate, Spring 2017
  • Explored incorporating conditional random fields into the implementation of the Fully Convolutional Network which uses only convolution and deconvolution layers in order to generate smooth pixel-wise labels for an image
  • Implemented the same in end-to-end fashion by transferring the learned representations of classification networks (viz. VGG net) to the segmentation task
• Latex Code Generation | Guide: Prof. Suyash Awate, Prof. Ajit Rajwade, Autumn 2016
  • Developed a module which generates the latex code for scanned images of printed latex equations
  • Used principles and techniques like adaptive thresholding with mean filtering, Hough transforms for image binarization, skew correction and hu-invariant moments
• Intelligent Pac-Man Agent | Guide: Prof. Shivaram Kalyanakrishnan, Spring 2017
  • Built a single and multi-agent intelligent agent which can play Pac-Man utilizing heuristic based Minimax and Expectimax search with Alpha-Beta pruning for tackling multiple ghosts
  • Performed inference using Particle Filters and Dynamic Bayes Nets for tracking the positions of the ghosts
• Flappy Bird Learning Agent | Guide: Prof. Ganesh Ramakrishnan, Spring 2016
  • Developed an AI agent for the Flappy Bird which outperforms an average human attempt
  • Used Neural Networks and Genetic Algorithms which simulated survival based on the principals of mutations, survival of fittest and cross-breeding to update the underlying Neural Network
• Readers Space | Guide: Prof. S. Sudarshan Autumn, 2016
  • Developed a platform for readers with auto-suggestion feature which uses Page-Rank type algorithm for book suggestion based on correlation between friends profile, mutual interests, and ones own history of literary activities
• Others:
  • Link Prediction in Social Graphs: Implemented a state-of-the-art link prediction algorithm in social network graphs in the SNAP framework.
  • Learning to Run (NIPS 2017): Explored Deep Deterministic Policy Gradients algorithm to synthesize physically and physiologically accurate motion of a human musculoskeletal model
  • C-like Compiler: Developed a compiler for a subset of C grammar which can perform dead code elimination on the intermediate code using live variable Analysis during compilation, for MIPS instruction set using flex and bison C++
  • Multi-Function Elevator Controller: Designed and implemented an elevator controller to handle concurrent request to minimize energy consumption using heuristic elevator scheduling algorithm in VHDL
  • Distributed password cracker: Built a distributed client-server system where client requests a hashed password to the server, and server divides the job dynamically among workers and returns the cracked password back to the client
  • Surveillance Bot: Designed a robot capable of shooting the desired target automatically using Edge Detection techniques and Control loop feedback mechanism on a microcontroller
  • Library Helpmate Bot: Fabricated a bot that can help curators in maintaining book clusters by using onboard book detection and shortest path line following algorithm in embedded C