I'm Prathmesh. I build — double-entry ledgers with strict multi-tenant isolation, a 20-plus-chain perpetuals exchange with 50x leverage and sub-second order streams, and indexers that have processed close to a billion on-chain events, — LangGraph research harnesses and RAG pipelines evaluated against live market feeds and retrieval benchmarks rather than static tests, plus an autonomous Telegram trading assistant reading RSI, MACD, and order flow, and — ingestion pipelines sustaining 120 million events a day across six schemas, sub-second Redis/Socket.io fan-out, and TimescaleDB architectures tuned to reclaim 2.4 terabytes across 379 compressed chunks.

Read this however you'd rather read it

Same information, three shapes. A recruiter skims the page, a terminal person runs curl, an agent reads the structured file.

Two years, roughly in order

Hover a point for what happened there. Or skip the line and just ask — these answers are grounded in the work below, not generated on the spot.

Answers appear here, with where they came from.

Selected builds and writing