TESA Miami: Skills, not connections
Track 03: Brain Gain · Hack for Humanity: Miami

The problem
Miami trains talent and then loses it. In a recent Gensler survey, 51.8% of Miamians aged 18 to 34 said they’re likely to leave, and Miami is now the second most expensive metro in the U.S. At the same time, South Florida employers need to fill more than 75,000 tech openings by 2034. The gap isn’t only jobs or skills. Graduates can’t see a path to a local job, especially first-gen students with no professional network.

What TESA does
Upload a resume and TESA shows which Miami jobs you’re closest to. It teaches the gap through questions, tests you with on-the-job simulations, and tells you when you’re ready to apply. Then you share a Skills Passport: your resume plus graded evidence of what you can do. It replaces the referral you never had.

How it works

Rank: your resume is mapped to O*NET skills and compared with 12 key Miami jobs, using Miami wages and job counts from BLS.
Gap: skills are ranked by importance × gap and shown on Bloom’s taxonomy, from Remember to Create.
Learn: a Socratic coach asks one question at a time and never gives the answer. It works in English, Spanish or Haitian Creole.
Simulate: three-step workplace scenarios at Apply, Analyze and Evaluate, graded against a rubric written up front.
Prove: at 85% readiness an alert fires, and you share a private, revocable Skills Passport.

How we use AI
Claude runs through the Cloudflare AI Gateway:

Claude Sonnet 5.5 reads resume PDFs directly, writes simulations, and grades answers at temperature 0.
Claude Fable 5.1 runs the Socratic coach and the mentor.

The models read, ask, generate, grade and explain. Code computes every number: readiness, gaps, wages and the ready alert. Resume claims are capped at Bloom level 3, so higher levels have to be shown live.

Who it’s for
Our persona is Andrés, a first-gen FIU computer science senior from Hialeah with an out-of-state offer and nobody to refer him. TESA shows him he’s already close to a Miami network support job ($78,910 median) and coaches him past the 85% readiness line. He stays.

Built with

Data: O*NET 31.0 and BLS OEWS
Server: Node and Express
Web: Vite and TypeScript, with a UX prototype built in Lovable
Knowledge graph: Cytoscape.js

Open source under the MIT license.