I build real-time systems, intelligent agents, and infrastructure that scales.
I'm currently focused on:
- ⚡ Real-time systems: websockets, pub/sub, distributed backends
- ♟️ A tiny open-source Lichess clone — local-first, scalable, and minimal
Recently released concall-parser, a python package to get structured features from earnings call reports. Made in collaboration with Jay Shah.
- Diving deep into how Lichess works — game servers, move validation, and scaling multiplayer chess
- Exploring ML inference infra — model servers, autoscaling, request batching
- Studying database internals — data modeling, indexing, replication
-
AI Research Assistant
Turn research papers into structured insights using local LLMs, vector search, and semantic chunking.
Tech: Python, Ollama, MongoDB, FastAPI -
Real-Time Leaderboard Engine
Matchmaking + scoring using Go, Redis, and Docker. Built for low-latency, high-concurrency use.
Tech: Go, Redis, Prometheus, Docker -
Multi-Agent Recommender System
Recommend users to users using OpenAI agents, MongoDB vector search, and intelligent prompts.
Tech: OpenAI API, Docker, AWS, FastAPI -
Forex Trading Bot (Currently undergoing refactoring, previous design was terrible) Streaming market data + trade signals using Python & GCP Pub/Sub.
Tech: Python, Pub/Sub, Docker, Dash -
Web Scraping (Multiple projects) Built scrapers to fetch jobs from indeed and YCombinator, to match against a resume reducing time spent searching for jobs. Tech: Python, Taipy, BeautifulSoup, Selenium
Others: Airflow, Github Actions, AWS, Pytorch
I write about building AI tools, scaling real-time systems, and going from toy projects to production-grade systems.