Vraj Patel
M.S. Data Science student at Stony Brook University. I build low-latency financial analytics engines, distributed ML pipelines, and scalable data-driven applications. Experienced in optimizing WebSocket streams and implementing real-time options strategies.
Technical Skills
Languages, frameworks, and tools I use to build scalable systems.
Featured Projects
A selection of my recent work in full-stack development, quantitative data engineering, and real-time systems.
Lot Lab – Urban Planning Simulator
Apr 2026Built an on-device urban planning simulator for NYC vacant lots on the Acer GN100 (NVIDIA GB10 Grace Blackwell Superchip), ingesting 7+ NYC open datasets and scoring parcels across 12 human & environmental use types with GPU-accelerated RAPIDS pipelines. Implemented NVIDIA cuOpt district-level optimization producing population-weighted plans; wrapped the platform as an OpenClaw skill with Nemotron-driven recommendations and a local Flux render pipeline for concept visuals.
Built a high-performance async data pipeline ingesting Polymarket order books at 182 snaps/sec across 35k+ active on-chain markets. Building a leading indicator engine leveraging crowd wisdom signal processing; conducting hypothesis testing on whether decentralized markets are immune to Goodhart’s Law.
Architected a geospatial clustering engine matching users by directional bearing, proximity, and time windows to dynamically calculate distance-weighted Uber fare splits. Integrated Google Gemini 2.5 Flash API for a context-aware chatbot and personalized CO2 reduction insights.
Hackathon-winning, production-grade event platform in talks for official university integration. Features a precision RAG chatbot using OpenAI gpt-4o over ChromaDB, automated AI pipelines with ~90% API cost reduction, and sub-200ms real-time sync via WebSockets.
Engineered a multimodal trading interface executing real-time NSE/BSE orders, proven to reduce latency by 60% and input errors by 80% in A/B testing. Integrates RAG for context-aware market sentiment analysis, demonstrating agentic AI workflow capabilities.
Work Experience
Building scalable infrastructure and extracting signals from noise.
FinTech Data Scientist Intern
Flits (India)
- Built real-time options analytics engine processing NSE/BSE tick data, implementing asyncio WebSocket client with 40% reduction in monitoring overhead.
- Engineered ETL pipeline parsing 170K-row instrument master file, applying SIC code mapping and outputting normalized PostgreSQL schema.
- Developed sector rotation tracker correlating NIFTY movements, implementing z-score normalization to identify top decile outperformers.
Real-time Options Analytics Engine Architecture
Data Analyst Intern
YHONK – Noise Pollution Mitigation
- Built distributed web scraping pipeline (Selenium Grid) with rotating proxies, extracting 50K+ school records with exponential backoff retry logic.
- Executed geospatial analysis on 2.6M GPS-tagged honking events, implementing PostGIS queries and R-tree indexing to compute violation hotspots.
- Applied ARIMA time-series decomposition on hourly violation counts, detecting 171% weekday anomaly and 130% seasonality.
Geospatial Analysis of 2.6M Honking Events
International Research Intern
AIT Brain Lab (Thailand)
- Fine-tuned T5-base and FactorSum (BART-based) transformers on ParaSCI dataset, achieving 0.42 ROUGE-L score for abstractive summarization.
- Built Flask REST API with Celery task queue for async inference, implementing request batching to reduce P95 latency from 4.2s to 1.8s.
Education
Academic background and degrees.
Certifications
Get In Touch
Have a question or want to work together? Drop me a message and I'll get back to you as soon as possible.