About
I’m Deepak Bansal - engineer, builder, and founder of Arcitae Labs.
I spend my time on production AI systems for FinTech and Legal Tech companies, particularly where audit, compliance, and reliability matter as much as the model output. RAG pipelines, multi-agent orchestration, LLM evaluation, prompt versioning - the infrastructure under the AI features that actually ship.
For the last decade, I’ve built and led engineering teams in regulated SaaS. At Klub, a revenue-based financing company in India and the UAE, I scaled engineering from 2 people to 70+ and led the platform from concept to over a billion dollars processed. I led ISO 27001 certification and ADGM/FSRA regulatory audits. Before that, I was on the founding engineering team at InCred, where I built the lending stack that now supports a $1.2B loan book.
In 2025, I led engineering at Lucio AI, a legal tech company shipping AI-assisted drafting, email workflows, and document processing. We shipped three AI products in six months, built our RAG pipeline on Turbopuffer after migrating off MongoDB, and stood up an LLM evaluation framework that caught hallucinations before users saw them.
That experience - production AI on regulated data, at startup pace - is the work I want to do for the next decade. Arcitae Labs is the consulting practice I started in 2026 to do that work for funded FinTech and Legal Tech companies, and occasionally for adjacent regulated categories where the same patterns apply.
I’m based in Bangalore, work primarily with US, UK, and Singapore companies, and reply to thoughtful emails within 24 hours.
Currently writing about production AI infrastructure, RAG at audit-grade quality, multi-agent orchestration in regulated environments, and the partner-economics dynamics emerging across incumbent SaaS platforms.