Software Engineer / AIML
Building production NLP pipelines at UW–Madison; published ML research with Springer.
Software engineer building Python backends with FastAPI and Flask, ML/NLP systems with PyTorch and BERT, and production deployments on AWS and Docker.
Connect on LinkedInLyntra
Wisconsin School of Business
StoccGuru
Languages
Backend & APIs
Data & Databases
Machine Learning
University of Wisconsin - Madison
M.S. Data Science · 2024 - 2026
University of Mumbai
B.Tech. AI & DS · 2020 - 2024
LightGBM and Gradient Boosting for Optimizing Shipment Mode in Pharmaceutical Supply Chains
Springer Singapore
Read PaperCurated work — @Tydos on GitHub
A deep learning system to identify derogatory tweets related to gender, race, age, religion, ethnicity, or sexual orientation. Built on a dataset of 110,000 tweets and evaluated using SVM, Random Forest, BERT, and an ensemble of BERT and LSTM models.
A deep learning based ISL recognition system using CNN-LSTM architectures. Data was captured via OpenCV, augmented, and processed using MediaPipe Holistic for feature extraction, achieving 85% accuracy. Includes Android deployment.