Founding AI/LLM Engineer at BlueBear AI — I build LLM agents that survive contact with production.
Most of my work is the unglamorous half of the problem: structured output that holds its shape, tool calls that fail loudly instead of silently, and guardrails that decide what an agent may touch before it touches it.
debawho.xyz · LinkedIn · debashishthakur9@gmail.com
Leading agent engineering at BlueBear AI — multi-agent workflows for B2B automation, MCP servers, structured-output pipelines, and skill- and rule-based access controls that constrain what an agent can do.
- Ayna AI — founding ML engineer. LLM/VLM analytics framework across 10K+ workflow outputs (cut error rates 18%); ComfyUI image pipeline serving 3K+ requests a day at 40% faster inference and 30% lower infra cost.
- Inspiron Info Inc — ML engineer. Revenue forecasting at 87% accuracy, an LLM-powered text-to-SQL engine, and LSTMs trained from scratch.
| LekhakAI | Chrome extension for multi-LLM inference from anywhere in the browser — one prompt, answers compared across providers. On the Chrome Web Store. · repo |
| The Prompt Craft | Prompt intelligence platform benchmarking ~6,000 Reddit-scraped prompts, with vector-embedding semantic search across the corpus. |
| meanpdf | Local-first PDF toolkit — merge, split, compress, convert, all in-process. Nothing is ever uploaded. 65-test suite. Private for now. |
If you are building something that leans on agents and you want it to hold up in production, write to me. I read everything.