I’m a builder driven by solving real friction. Rather than building for theory, I engineer solutions for immediate impact, whether that means automating intricate desktop workflows, orchestrating sub-second vector retrieval across massive datasets, or crafting ultra-lean serverless platforms with zero overhead. Every system I write is focused on speed, resilience, and clean, reliable execution.
1st place winner out of 100+ national teams in a high-intensity 24-hr sprint. Architected an end-to-end automated business analytics platform recognized for speed, reliability, and pragmatic architecture.
Built OS-level visual scrapers using PyAutoGUI & OpenCV bypassing bot detection. Orchestrated async Django daemons and integrated Google Gemini Flash with strict JSON schemas, plus <250ms pgvector RAG.
Engineered scalable ETL data pipelines and REST APIs using Python & SQL to process, clean, and standardize high-volume datasets. Optimized database schemas and automated cloud storage ingestion with high fault tolerance.
Built zero-egress serverless event ticketing on Cloudflare Workers (<50ms admission check) with in-browser QR generation. Architected LiteLLM multi-provider gateway balancing OpenAI, Claude, and DeepSeek to cut API costs.
Interactive systems showcase — hover cards to lift from deck and inspect architectures, latencies, and production stacks.
Dual-retrieval enterprise RAG pipeline combining semantic pgvector search (HNSW index) and BM25 search with PostgreSQL multi-tenant RLS.
Multi-threaded Maps & web crawler scoring digital maturity + Thompson Sampling bandit algorithm dynamically prioritizing highest-yield outreach.
5GB+ Yelp review lakehouse pipeline with AWS S3 staging, distributed Snowflake transformations, and Python sentiment classification UDFs.
Telegram conversational assistant parsing unstructured income/expense entries with real-time budget telemetry, limits, and web dashboard.