Technical Program Manager

I build training and
certification programs.
These are my AI experiments.

Ten years designing certification programs at Google and TikTok. These prototypes explore what AI can do in the domain I know best.

See the experiments → j3146596@gmail.com ↗

The Experiments

3 AI prototypes I built to explore my domain.

Early-stage work. Not production systems — just me testing what's possible.

AI Experiment Hosted on Railway

Exam Blueprint Agent

An attempt to automate the exam design process I've run manually for years.

14 AI Agents
5 Human Gates
22.5h Build Time

I've run exam design workshops and psychometric reviews professionally at TikTok and Google. This pipeline uses 14 AI agents to replicate that process — from evidence gathering and SME simulation to bias review and blueprint generation — with those standards in mind, not as a certified claim. Five human-in-the-loop gates keep a person in the decision chain. Built as a personal experiment to see how far AI can go in a domain I know well.

Node.js TypeScript Claude API Multi-model Supabase SSE Railway
Open Experiment ↗
AI Experiment Hosted on Vercel

Candidate Experience Portal

A minimal candidate portal for the kind of certification program I've been managing since 2016.

18 E2E Tests
RLS DB Security
✦ Magic Links

Having managed certification portals at Google for Education and TikTok Academy, I wanted to see how quickly I could prototype one from scratch using AI. This covers the candidate lifecycle: auth, eligibility tracking, prerequisite validation, and cross-domain exam handoff. Early-stage — not designed for scale.

Next.js 16 React 19 TypeScript Tailwind 4 Supabase Playwright
Open Experiment ↗
AI Experiment Hosted on Vercel

Agentic Skills Lab

An AI-graded hands-on exam environment — the kind of assessment I've designed manually for years.

4D AI Rubric
Auto Codespaces
2× Score Tracks

Performance-based exams are the hardest part of certification design to get right. This prototype provisions an isolated coding environment per candidate, runs deterministic validation, then uses a four-dimension AI rubric — Delegation, Description, Discernment, Diligence — to assess how well they worked with AI. An experiment in whether AI can grade what humans currently review.

Next.js 15 TypeScript GitHub API Claude API Supabase Vercel
Restricted Access

Background

What I actually do.

Certifications & L&D

Exam development & JTA facilitation
Psychometrics & item writing
Credentialing & digital badging
ANAB / ISO 17024 program design
Learning platform strategy (Docebo, Intellum)

Program Management

0-to-1 product and program launch
Cross-functional stakeholder management
Vendor and BPO operations
Data-driven program iteration
Agile / Scrum facilitation

AI Experimentation

Anthropic Claude API
Multi-agent workflow design
Human-in-the-loop systems
Supabase · Vercel · Railway
Prompt engineering & iteration

Data & Platforms

SQL / BigQuery / Data Studio
ETL pipeline design
Salesforce / CRM operations
Google Analytics / GTM
LMS administration

Get in touch

Let's work on something real.

Open to TPM roles in AI, EdTech, certifications, L&D, and eCommerce platforms.

j3146596@gmail.com ↗ LinkedIn ↗ GitHub ↗