What I've Built

Ongoing engagement with an investment firm where I've built out their research infrastructure from scratch. Started by understanding how their analysts actually work, then built LLM-powered tools that generate research reports at multiple depths. The work has expanded into data infrastructure, portfolio analytics, and a unified dashboard that brings together company research, screening, and portfolio views in one place. I work directly with the analysts and iterate based on how they actually use the tools.

I built and deployed a production ML model for a logistics technology company to optimize their scheduling operations. I used historical data to train an ensemble model that predicts optimal time slots, ran comprehensive back-testing to validate performance, and deployed an inference endpoint that integrates directly into their application for real-time predictions.

I designed and delivered an all-day applied AI workshop for faculty at a large university hospital. I worked with the organization's leadership to understand their specific pain points, then built a custom curriculum focused on practical AI applications for healthcare settings. The workshop included hands-on exercises and I provided follow-up resources including instructional videos.

Started as a technical assessment of a healthcare technology company's document generation platform, then evolved into an ongoing engagement building the system. I'm designing extraction and generation pipelines for complex regulatory documents where accuracy on structured data has to be near-perfect. The approach combines agentic extraction — where the AI reasons across documents and adapts to variability — with deterministic generation so the output is reproducible and auditable. I'm also building the ground truth testing infrastructure and working closely with the engineering team on architecture decisions.

Most teams have AI tools but no shared way of using them, and nothing connecting what engineers are doing back to the rest of the organization. I work with the team over weeks to build the system around the tools: project documentation that gives AI real context about the codebase and domain, conventions for how to plan, build, and review AI-assisted work, and workflows that connect engineering output to product, design, and stakeholders. The important part is that each week's work feeds into the next. Docs get better, conventions get sharper, the team gets faster. By the end the whole process is theirs to keep running and improving.