Hi, I'm
Ganesh S
AI Researcher. ML Engineer.
About me
.
I'm a final year undergrad from IIT Madras, pursuing a dual degree with Masters in Data Science and Bachelors in Mechanical Engineering. I have a deep passion for research and appliation in AI and Machine Learning.
Experience
.
AI Research Intern - Fujitsu Research
Worked on question-answering systems using LLMs for table-text sources.
AI Engineer Intern - Sarvam.ai
Worked on LLM agents. Benchmarked and tested various open souce LLMs on their function-calling abilities.
Software Developer Intern - Standard Chartered
Worked alongside the Data Analytics Team at Standard Chartered on a migration automation project.
Research
.
Master Thesis - Deployability of LLM designed Rewards for Restless Multi Armed Bandits
Advisor: Prof. Balaraman Ravindran
Collaborating with Google DeepMind (Prof. Milind Tambe and Dr. Aparna Taneja) to analyze impact of multinguality on fairness and performance of LLMs in the context of bandit problems.
Publication under review at an A* conference workshop.
Fujitsu Research - Question-answering systems using LLMs
Guides: Ankush Agarwal, Pranoy Panda and Chaitanya Devaguptapu
Worked on increasing efficiency and accuracy of question-answering systems using LLMs for table-text sources.
Publication under review at an A* workshop.
Projects
.
Cloud Vault
NextJS
FastAPI
DynamoDB
AWS
PyTorch
Fully cloud-native solutions with features including fraud monitoring and customized underwriting systems.
Appian AI Challenge
NextJS
FastAPI
Llama Index
OpenAI
BPM
Generative-AI based Business Process Management for Health Management of Corporate Clients.
InstiSpace
Flutter
NestJS
GraphQL
PostgreSQL
AWS
Community mobile applications for students of IITM with 5k+ active users.
Investor Insights
NextJS
PyTorch
NLTK
Tableau
Application for institutional investors to monitor stocks, gain insights and find the latest news.
BTC/USDT Algorithmic Trading
Technical Analysis
Data Science
Python
An explainable ML based Algorithmic Trading Model for trading BTC/USDT in the cryptocurrency market.
Facial Keypoint Identification using ML Resnet
PyTorch
CNN
Kaggle
A Masked Loss based Residual Convolutional Neural Network model, to identify the main keypoints in a face. Despite the dataset lacking a lot of labelled keypoints, the model was able to attain an impressive RMSE score of 2.20