Getting an AI agent working in a demo is easy. Getting it reliable, observable, secure, and affordable in production is where most projects stall.
I build agent systems and the infrastructure around them: at Speakeasy on the AI Control Plane and MCP tooling, and in open source with AgentLink. For clients, I design and build agents, MCP servers, and LLM features that are ready for real users.
Who it's for
- Product teams adding LLM features or agents to an existing product
- Companies that want to expose their APIs and data to AI agents through MCP
- Teams with an AI prototype that needs to become production ready
What you get
- Agents and workflows with clear tool boundaries and evaluation
- MCP servers that expose your systems safely, with proper authentication
- Monitoring, cost controls, and fallbacks for LLM features
- Clean, maintainable Golang or TypeScript code your team can own
How I can help
When you need AI systems built, not just advice. Production-grade agents, integrations, and platforms.
AI Agents & Workflows
Tool-using agents and automated workflows designed to be reliable, observable, and cost aware.
MCP Servers & Integrations
Expose your APIs, data, and internal tools to AI agents safely through the Model Context Protocol.
LLM Features in Production
Bring LLM features from prototype to production with the monitoring, security, and fallbacks they need.
API & Platform Engineering
Golang and TypeScript backends, OpenAPI-driven APIs, and cloud platforms on AWS and GCP.
Ways to work together
Advisory Retainer
A few hours a month of on-call advice, reviews, and decision support.
Enablement Sprint
A focused engagement of a few weeks to roll out AI tooling and workflows across a team.
Fractional Leadership
One or two days a week as your CTO or engineering lead, embedded with your team.
Project Delivery
Scoped builds of agents, MCP servers, and AI features, delivered end to end.
Frequently asked questions
What is an MCP server?
MCP (Model Context Protocol) is an open standard for connecting AI agents to tools and data. An MCP server exposes your APIs, databases, or internal tools so agents like Claude, ChatGPT, and Cursor can use them, with access you control.
Which languages and platforms do you build with?
Primarily Golang and TypeScript, deployed on AWS or GCP, using models from Anthropic, OpenAI, and other providers depending on the job.
Can you take over an existing AI prototype?
Yes. A common engagement starts with a short review of the prototype, followed by hardening it for production: evaluation, monitoring, security, and cost controls.
Do you offer ongoing support after a build?
Yes, through an advisory retainer or follow-on project work, so your team has support as the system evolves.
What Clients Say
Reviews from founders and teams I've worked with.
Tristan helped us work through some of the challenges in setting up and launching an API, including it's integration into our stack, and provided a comprehensive review of our product which has enabled us to move forward. Tristan was always a delight to work with; professional and friendly and quick to reply, and he made sure I understood the technicalities of what he was proposing. Would highly recommend working with Tristan.
Lucy Sattler
API launch and product review
Consulted Tristan for general startup software engineering advice and direction for my preventative health business. He brought a wealth of knowledge and clearly laid out my options in a pragmatic, easy-to-understand way, which made the decision-making process much clearer.
Christopher Collins
Startup technical advice
Let's talk about where you are
Based in Cairns, working remotely with teams worldwide. Tell me what you're working on and we'll work out whether I can help.
Get in touchRelated services