Adam M.

I'm Adam, a Senior AI Engineer, and I build production AI systems that businesses rely on every day: AI agents, RAG pipelines and LLM workflow automation, mostly for companies where a wrong answer costs real money.

I've been doing this on Upwork for seven years. I'm Expert-Vetted (Upwork's top 1%, picked by interview rather than by algorithm) and I've kept a 100% Job Success Score across 75+ projects.

WHAT YOU CAN EXPECT

• Systems that keep working after launch. Everything I build ships with schema validation, evals and a regression suite, so six months later it still does the job and you can prove it.

• It plugs into what you already have, whether that's your CRM, your database, your APIs or your cloud. I'm AWS certified (ML Specialty) and equally happy on GCP.

• You'll always know where things are. I pride myself on excellent client communication, so you'd get an update from me every day at the very least, plus a weekly call or Loom where I walk you through the progress. Clients tell me this is the part they remember most.

SOME RESULTS

• On an AI underwriting platform used by 15+ insurers across the US, policy review went from days to 3 to 5 minutes, at 99.55% precision on the rules that move money.

• A UK mortgage broker cut manual payslip review by 80%. Two models read every payslip and only the ones they disagree on go to a person.

• A UK government-data client got 12 million planning documents extracted, embedded and searchable in under 48 hours, with self-hosted embeddings, so there's no per-token bill.

• I built the assistant inside a UK consumer money app: seven specialised agents and 43 deterministic tools, with every financial figure worked out by code rather than the model, so it can't make up a number.

Most of my work turns into long-term engagements. The first project usually automates one workflow, and then we get on with the rest.

WHAT I DO

1. AI agents and workflow automation. Multi-agent systems that take a manual process from start to finish and make real decisions along the way. I build them in LangGraph and Claude with MCP, schema-validated and observable at every step, with a person in the loop wherever one belongs.

2. RAG and LLM engineering. Retrieval pipelines, knowledge assistants and chatbots over your own data, plus cost and accuracy work on LLM systems already in production. One pipeline I built summarises 700+ UK planning policy documents every week, fully automated, at about 75% lower LLM cost thanks to context caching on Gemini 1.5 Pro.