Justin Hawkes

I build production AI systems that make real operations measurably smarter, from architecture to deployment. I shipped a production AI operations platform solo on Azure, delivering cited-answer retrieval over the company’s own documents, a deterministic ingestion pipeline, and an observability layer.

I also built Nova, my own multi-agent assistant, and I’m deliberate about building for the friction people actually hit. My work follows a forward-deployed pattern: embed with operations and sales teams, find the workflow that slows people down, build the system that speeds it up, and ship it to the people doing the work.

Before I focused on AI, I delivered measurable operational impact through engineering and tooling. I built a parametric AutoCAD/VBA program that turned an 8-hour drafting task into about 30 seconds, and I created an Excel-based ordering tool adopted company-wide that saved hours per person per week.

I’m drawn to AI that proves itself in practice: I built Model-Bakeoff, an LLM evaluation harness, and Job-Radar, a multi-source job-discovery engine with retrieval scoring at scale. I bring that same engineering mindset—deterministic pipelines, observability, and real-world constraints—into every system I ship.