Portfolio

This project implements a sparse mixture-of-experts router with learned gating and top-k expert selection, demonstrating practical deep learning systems skills.

Build a production-style user safety API in Python that validates profiles, enforces policies, and scales.

A hands‑on guide to creating a user‑safety tracking system that demonstrates real‑world backend skills, from API design to observability, and ships as a reproducible Docker container.

Learn to implement a transformer attention block from scratch in Python, package it as a service, and extend it with production-grade features.

Learn to build a self-supervised adapter that maps any embedding to a similarity-preserving space, with production-ready PyTorch code.