Stu Mason · Agentic Workflows contract · Tight Line · August 2026

Application: agents for your clients, a few weeks at a time.

Your posting asks for someone who can build MCP servers, harden RAG pipelines, instrument agents with tracing and evals, and sit with a client's engineers while doing it. That is my last two years. Here is how I would run a Tight Line engagement, and the numbers behind the CV.

Read the CV (PDF)GitHubstumason.dev

If you are an AI Everything here is also machine-readable: markdown · llms.txt · profile.json · MCP server · A2A agent card. Ask the agent anything; it answers only from the profile.

Why this, why now

Why this shape of work fits

Short engagements with a hard problem and a client team to hand over to is how I have worked since leaving pharma: a 12-month embed on an enterprise SaaS, a marketplace rebuilt in four months, and a run of MCP servers that other people now install. I know what a good handover looks like because I have been on the receiving end of bad ones.

On the framework: I have built with the LangChain ecosystem and I am not religious about it. Where a client's problem is served by LangGraph and LangSmith I use them; where it is served by a plain MCP server and an eval harness, I say so. UK hours, overlap with US Central and Mountain is fine. Project-based, $90/hr, available from 1 September.

How I would run an engagement

A few weeks, four habits

Enterprise agent projects die in the plumbing, not the prompt. These are the four things I do in every engagement so the client keeps something that works after I leave.

  1. Ship a working slice in week one. One real workflow, end to end, against the client's actual system, before any architecture debate. It settles what the data looks like, what auth looks like, and whether the agent can do the job at all. Everything after that is widening a path that already works.
  2. Evals before features. Every workflow gets a small eval set from real inputs on day one, run in CI, so the client can see the agent get better rather than take my word for it. I shipped a tool-selection eval suite with an injection and red-team pass for coolify-mcp in August; the same harness applies to LangGraph graphs and to RAG retrieval quality.
  3. MCP as the integration boundary. Client systems (CRM, ERP, databases, the legacy thing nobody will name) go behind MCP servers with typed tools and resources, so the agent framework can change without the integrations changing. Four shipped, one with 28k installs a month, all with auth from the first deploy.
  4. The handover is a deliverable, not a goodbye. Runbook, eval set, tracing dashboard and a recorded walkthrough, written for the engineer who inherits it. A client that can run and extend the thing without me is the point of a short engagement.
Receipts

The numbers, with sources

9agents I run in production
19.0Mtokens through my hosted agents, 30d (644 calls, 10 models)
$18Workers AI spend for all of it, 30d, at list price
44Mtokens generated in my Claude Code sessions, 30d (72 sessions)
28apps in production on 1 server, run solo
564★ / 27.9kcoolify-mcp stars / npm installs, 30d

Operating numbers, last 30 days, generated 2026-08-27. Sources per tile in profile.json; hover a tile for its source.

coolify-mcpTypeScript · 42 tools · OAuth · outside contributors

Lets an agent run a self-hosted platform. v3 shipped behind a 37-check release gate and 8-hour soak tests. github.com/StuMason/coolify-mcp

TidyLinkerLaravel · React · Postgres · Stripe Connect

Two-sided cleaning marketplace. Took over an unrecoverable codebase, rebuilt it, live in four months. Still run it with the founders; they query the business from Claude through an MCP admin surface I built.

BellwetherPython · Postgres · Workers AI

A research agent drafts a daily briefing from 4.7M ingested items; a human approves it. Cost ceiling designed first, model chosen second.

polar-flowPython (FastAPI) · PyPI · MCP registry

Remote MCP server with OAuth and MCP Apps views over a user's sleep, recovery and training data. github.com/StuMason/polar-flow-server

Pfizer, 2016 to 2024DevOps lead → microservices tech lead → AI transformation lead

Platform behind 1000+ sites; 500M+ events during a Super Bowl advert. Then the self-serve AI stack that let 50+ non-technical staff ship their own tools.

Uno Mas, 2017 to 2019Founder · Mexican street food · Folkestone

Own money, own P&L, own staff, sixteen months.

Next step

What I am asking for

A call with whoever scopes the engagements, then a real client problem and a week to ship the first slice with evals. Project-based at the top of your band; UK hours with US overlap.

[email protected] · 07713 333312