AI Engineer · ships LLM systems
Gopal
Bagaswar
I build LLM systems end‑to‑end — agents, RAG, voice AI, with tests. Five years reading the data; now I build the systems that do.
- 2 apps live on the App Store
- merged PRs · mem0 · Gradio · LiveKit
- 5 yrs analytics → AI
Remote · IST (3–4h EU + US‑ET overlap) · open to relocation, needs visa sponsorship
01 — About

I build the systems / I used to just measure.
Five years reading the numbers. Now I build the systems that produce them.
LLM agents, RAG, evals — with tests.
Put an app on the App Store people actually use.
Bengaluru, IST. Remote-ready worldwide, open to relocation with visa sponsorship.
02 — The map
Analyst → engineer
Four roles, one MS, one direction.
Aug 2021
Data Analyst
TCS · Hyderabad
Prophet / ARIMA forecasting, automated dashboards. TCS iGnite grad
Aug 2024
Analyst
Nielsen · Mumbai
Audio analytics for 100+ radio stations, QA pipelines. 100+ stations
Jun 2025
Senior Data Analyst
Landmark Group · Bengaluru
Shukran loyalty — 10M+ members, 6 GCC markets. Dubai-HQ retail
Nov 2025
Senior AI / Data Analyst
Catalyst Brands · Bengaluru
60× automation — now bringing LLM + RAG into the workflow. formerly JCPenney
→ Next
AI Engineer
remote · worldwide
The whole point of this site. the only node not true yet
parallel · 2022 – 2024
MS Computer Science (AI/ML)
Woolf University
HuggingFace Agents + MCP · Karpathy Zero-to-Hero · a self-designed 180-day curriculum.
03 — Selected work
Things that ship
Everything below runs. Hover any tile for the story.
01 / 02scroll to explore ↓

The command center for my AI-engineer hunt — a job-radar engine auto-applies, the dashboard tracks every move.
Next.js · Python job-radar engine · 6 ATS APIs · 65 tests green
Solo · my own job hunt
Career OS
LIVE · 65 TESTS
PAPER-ONLY · NO LIVE CAPITAL
An autonomous crypto trading system — 8 strategies, walk-forward tested, paper-only.
Python · event-driven backtester · walk-forward validation · risk engine · Hyperliquid perps
Solo · research only
HedgeBot
974 TESTS · 8 STRATEGIES
6 DEPARTMENTS · PRIVATE
My chief-of-staff agent system — 90+ agents across 6 departments, always on.
Python 3.12 · Claude SDK · 18+ MCP servers · launchd · SQLite
Solo · 24/7
MAX HQ
90+ AGENTS · 6 DEPTS
Ask me about any of it →
04 — Open source
Merged, not drafted
Merged into the tools I actually use. Code, not docs.
01 / 02five, all merged ↓
05 — The stack
- LLM systems
- Multi-agent orchestrationRAG (hybrid BM25 + semantic, re-ranking)MCP / FastMCPLangGraphPydantic AIClaude Agent SDKtool callingagent memory
- Evals & reliability
- DeepEvalLangSmithPromptfoostructured outputsconfidence-gated abstentionCitations-grounded responsesprompt-injection defenses
- Languages & backend
- Python 3.12 (async, Pydantic)TypeScriptFastAPINode.jsSQLBigQueryPostgreSQLpgvectorChromaDB
- Product & infra
- Next.js (App Router, RSC)ReactTailwindCSSTwilio VoiceStripeDockerGitHub ActionsAWS BedrockVercelCloudflare Tunnel
06 — Say the word
You read this far. We should probably talk.
gopalbagaswar19Bengaluru / IST, with 3–4 hr EU + US-ET overlap. Open to relocation with visa sponsorship.




