Case study

PeptPal

PeptPal is a React Native + Expo app backed by FastAPI and PostgreSQL. It tracks inventory, doses, cycles, and biomarkers, while flagging dangerous protocols by comparing user weight against trial cohorts. The core wedge is an evidence engine with trust-tiered citations, harmonic-decayed weighting, and per-peptide hard ceilings.

RoleFull-stack engineering, mobile development, data modeling
TimelineOngoing
FocusHealth + Harm Reduction Mobile App
Mobile app developmentEvidence engine + dose scalingPharmacokinetic modelingCommunity consensus weighting
Expo + FastAPI + PostgreSQLStack
114 passing (Vitest)Tests

The brief

Challenge

Peptide forums propagate dangerous default doses copied from trial cohorts that are 60+ lb heavier than typical users, with no per-kg scaling and no source quality signal.

Approach

What we made

Built a per-kg dose scaling layer over a trust-tiered evidence engine, added a first-order degradation model so live vial potency is part of every injection decision, and wrote a weighted forum consensus that promotes bloodwork-backed posts over anecdote.

  • Evidence engine spans 17 peptides x 4 personas with trust-tiered citations (A-F).
  • Weight-scales every protocol dose and flags 'dangerous' at 1.4x trial per-kg exposure.
  • Per-peptide first-order degradation model with live potency bar and dose compensation math.
  • Weighted-median forum consensus: bloodwork-attached posts count 5x, 60+ day logs 2x.
  • Passphrase-encrypted JSON backup with no server-side storage of user data.

Outcome

Results

A working monorepo mobile app with a functional evidence engine, degradation tracking, biomarker panels, and a tested consensus pipeline.

React NativeExpoTypeScriptNativeWindFastAPIPostgreSQLTortoise ORMTurborepo

Gallery

Visual snapshots

Click any image to expand.

Next project

IMM Sourcing Pipeline

A four-stage Python pipeline that pulls injection molding machine quotes from Gmail, downloads supplier PDFs, extracts pricing, and builds a unified comparison document.

View next project