Black UX Labs: AI Career Discovery for 500 Participants

Black UX Labs is a UX research organization focused on career discovery and job search for Black professionals. I led product and engineering for their AI-powered career guidance platform, which was used in a 500-participant research study.

Traditional job search tools are narrow. They return job listings based on keyword matches, but miss the broader landscape of opportunities that could actually move someone's career forward: grants, upskilling programs, fellowships, and non-traditional paths. Many participants also struggled to articulate what they were looking for in a way that search engines could work with.

I designed and built a multi-stage AI agent pipeline using FastAPI and the Instructor library, orchestrating multiple models to handle different parts of the workflow.

Resume Analysis & Scoring The system ingested resumes and scored them against Black UX Labs' proprietary rubric, giving participants structured feedback on where they stood.

Query Generation from Resumes Using small function-calling models on Fireworks AI, the system converted resume data into structured search queries. This was the key insight: instead of asking users what they wanted (which many couldn't articulate precisely), the agent inferred the right things to search for based on their background.

Broad Opportunity Discovery Those generated queries fed into Exa AI with customized prompts designed to surface opportunities beyond traditional job listings. The system returned grants, upskilling programs, career pivots, and non-obvious paths that a standard job board would never surface.

LLM Orchestration The pipeline combined Fireworks AI Llama models for fast, cost-effective function calling with GPT-4 for higher-reasoning tasks, routing each step to the right model for the job.

I wasn't in this video, but it covers the Black UX Labs project and shares insights from the research:

Powered a 500-participant AI career discovery study that validated demand for agent-based career guidance

Achieved a passing satisfaction rate across the study cohort

Won the Goodie Nation x Google for Startups workforce grant based on the platform's impact

Built one of the earlier production AI agent systems (2023-2024), well before agentic workflows became mainstream