Hi, I'm
I build ML models and systems-level code from scratch. Specializing in Machine Learning, Transformers & NLP, and Systems Programming. I implement complex algorithms end-to-end: Autograds, RBF SVMs, RL models, and LMs. Strong foundation in OS internals and Computer Architecture.
A mix of competition-winning builds, tooling, and experiments.
Implemented Custom Model Based Linear-RL from scratch for inverted-pendulum system. The RL model maintains desired state with ±3% margin of error in angle. Gaussian supervised model for next state prediction achieves 98.7% accuracy on synthetic simulator data.
Designed and implemented a custom Autograd engine from scratch with secondary data structures and complex operations support. Built multi-layered linear regression, softmax, and sigmoid functions. Complete automatic differentiation implementation.
Designed and implemented Custom RBF Kernel SVM for breast cancer detection achieving ~97.6% accuracy with only 0.3% false negative rate (Type-II Error). Outperformed standard Neural Network trained on 128 iterations with 126 samples, 30 input dimensions.
Selected client work and independent engagements outside of academic projects.
Architected and delivered a full-stack Ruby on Rails web application — from database schema design to front-end integration. Implemented RESTful APIs, Devise authentication, and a responsive UI.
Deliverables
Stack
Designed and built a production-ready native Android app from scratch in Java. Covered UI/UX in XML, REST API integration, offline-first SQLite persistence, and Play Store submission support.
Deliverables
Stack
Built a backend service integrating with the GitHub API via Octokit — aggregating repo data, language stats, and project highlights. Implemented smart file-based caching with a 24-hour refresh strategy.
Deliverables
Stack
Delivered structured Android development sessions as App Dev Lead at Google Developer Group IIIT Kalyani. Covered the full Android lifecycle for a cohort of student developers — layouts to networking.
Deliverables
Stack
Open to freelance opportunities
Rails backends, Android apps, API integrations & more.
Google Developer Group IIITK
Build scalable native apps and run intensive technical sessions for student developers within the GDG IIIT Kalyani community. Responsible for curriculum design, hands-on code reviews, and mentorship for 80+ participants in technical workshops.
Led IronLegions to the semi-finals of Microsoft ImagineCup by building Course-Craft, an AI-powered educational platform for course and assignment management.
IIIT Kalyani, West Bengal
Mothers' International Academy
Carmel School