← All work

Hackathon winner · Genomics · Governed AI

MERIDIAN

A privacy-first pharmacogenomics system designed to keep medical AI inside explicit evidence and governance boundaries.

ROLE
Lead contributor on a three person hackathon team
STATE
Hackathon winner and systems prototype
Jack’s pitch · Stone & Chalk

Build

WHAT IT DOES

A governed genome pipeline, evidence retrieval, constrained model and source first interface.

HARD PART

Turning sensitive biological input into useful guidance without allowing the model to invent clinical certainty.

RESULT

A source bound pharmacogenomics prototype with explicit failure paths.

NOW
The app, control API, PharmCAT worker and deterministic clinical engine are implemented.
NEXT
Deploy the private processing path and validate it with deidentified data.

System flow

  1. 01GENOME INPUT
  2. 02GOVERNED PROCESSING
  3. 03EVIDENCE RETRIEVAL
  4. 04CONSTRAINED MODEL
  5. 05GUIDANCE