# Client AI delivery

> RAG and document Q&A systems scoped, built, and deployed inside clients' environments, with LLM-driven engineering automation alongside.

2024 → now · forward-deployed · RAG in clients' environments · LLM engineering automation · rag llm python fastapi kubernetes

Arc10's client work is where the way I work was set, and where it gets tested. I am the engineer the client works with directly: I scope the problem, own the architecture, write the code, and ship it into their environment. Engagements are not named here.

## The engagement

It starts with discovery: sitting with engineering, product, and non-technical stakeholders, finding the real constraint, and writing it down before anyone codes. Ambiguous requirements become a deployable system with a clear boundary, and the hardest slices get built hands-on rather than delegated.

## What gets deployed

Retrieval-augmented generation and document Q&A systems that run inside the client's environment: grounded answers over their own documents, delivered where the data already lives. The platform work on Arclyx and the client deployments feed each other, since the same problems of grounding, citation, and isolation show up in both.

## Engineering automation

Alongside the product work, LLM-driven automation for the client's own engineering process: issue triage, code review, and test generation, wired into how their team already ships.

## Why it matters for a forward-deployed role

This is the forward-deployed pattern in practice: one engineer accountable for architecture, delivery, and what happens in production, working in the client's environment, then the next slice.

Back: https://jigarpatel.ca/#work