Back to all posts

Building AI agents for an auto-transport brokerage

How I grew from frontend to full-stack to shipping AI agents in production.

Building AI agents for an auto-transport brokerage
Tech stack
AI AgentsLLMRAGPrompt EngineeringFull-Stack

My ongoing work is building AI agents and LLM-powered automation for an auto-transport brokerage platform. This is a high-level look at the skills behind that work — not the internal product — and the path that got me here.

The growth arc

I joined as a frontend engineer. As the product grew, so did my scope: I took on backend services, then DevOps and deployment, and most recently moved into AI engineering. Necessity, not a title, drove each step — and that's exactly why the growth stuck.

What I work on now

AI agents and LLM-powered automation: designing agent behavior, wiring tool-calling, applying retrieval (RAG) where it helps, and engineering prompts that are reliable in production rather than just clever in a demo.

Lessons that generalize

The model is the easy 20%; the engineering around it is the other 80% — tools, guardrails, evaluation and verifying output instead of trusting it. A tightly scoped prompt beats a clever one. And shipping AI responsibly means designing for the cases where the model is wrong.

Note on confidentiality

This write-up intentionally stays at the level of skills and general lessons. It contains no internal architecture, data, features or roadmap — only what is safe to share publicly.