SVC · 01
Measurement & job architecture
SIOP job analysis, competency models, leveling — the spec your models should follow.
Taxonomies · rubrics · automation-readiness scores · HRBP/legal docs
Applied ML for workforce strategy.
MLOps and neural networks for people analytics — built so leadership can make workforce decisions backed by real data, not gut feel.
Applied Machine Learning Engineer focused on people analytics and workforce strategy. People Research Scientist at Meta. M.S. I-O (SJSU), B.S. Psychology (CofC). Agents since Nov '22; neural nets on comp and attrition at Aon before that. Some consulting on the side.
Practice
Usually three teams: I-O writes the rubric, data science trains the model, MLOps ships and monitors it in production. I do all three, which matters when the work has to hang together.
- Pre-2022Neural nets for job classification, attrition, pay equity — Aon, then Meta
- Nov '22 →Agent pipelines for job analysis and people research
- NowContract work: psychometrics, MLOps pipelines, agent orchestration for people analytics and workforce strategy teams
Services
Fixed scope. Running code and documentation, not a strategy deck.
SVC · 02
Skills & inference
Ontology design, validation protocols, match-rate benchmarks.
Validation studies · governance rules · mobility integrations
SVC · 03
Agent pipelines
Multi-step workflows for people research — taxonomies, evidence, eval loops. In production since Nov '22.
Agent specs · eval harnesses · human review gates · tool contracts
SVC · 04
Neural nets & MLOps
Neural nets or classical ML, shipped and monitored in production — training pipelines, Python/SQL backends, React/Next if analysts need a UI.
Models · training pipelines · monitoring · analyst UIs · PyTorch / Spark
Engagements
Scope
Problem, stakeholders, data on hand, what done means for legal and the business.
Diagnostic
Read what you have, talk to users, write up gaps. Fixed fee.
Build
Weekly checkpoints. Methods written down.
Handoff
Artifacts, replication notes, walkthrough with your team.
Track record
From full-time roles building people analytics and workforce strategy systems. Same work I take on contract.
Nov '22
Agent systems for people research
1,227
Roles in job architecture platform
96%
Skills inference match rate
200+
Research workflows
People Research Scientist · Meta
- Job analysis platform — multi-agent pipeline across 1,227 roles (task taxonomies, competencies, automation-readiness)
- Skills intelligence — 70k employees × 20k skills, 96% match
- AI Adoption Index — CFA + bootstrap replication, 526 business units
- React/Next tooling for ~50 analysts; 200+ workflows
Earlier · Aon, GPTW
2014 — 2020
- Aon — neural nets on a global comp database; attrition, pay equity, high-performer models
- Great Place To Work — Fortune 100 Best Companies assessments
Education
M.S., I-O Psychology — San Jose State University
2015
B.S., Psychology — College of Charleston
2012
Stack
People analytics
Workforce strategy
MLOps
Neural networks
Multi-agent systems
SIOP job analysis
IRT / CFA / EFA
Skills ontologies
PyTorch
Python
SQL
React / Next.js
Spark
Lab
Public GitHub. Side projects and prototypes. Mobile-first, zero tracking, era-honest.
Vector Hoops
Daily NBA chimera puzzle — 12k player-seasons, era-honest stats, six ranked modes. Mobile-first fluid UI.
Vector Pitch
World Cup chimera — 633 tournaments, per-90 z-scores, attack+duel fusion. Same MTNN pattern, dark pitch theme.
Vector Gridiron
Your fantasy season, on rails — ESPN/Sleeper sync, model-backed start/sit, waiver & trade calls grounded in NFL embeddings. Mobile-first cockpit.
Vector Equities
Public companies as vectors — SEC EDGAR XBRL 2741 FYs, 283 tickers, 17 towers, 64-d transformer MTNN val 88.2% test 86.2% CQS 0.635. Same pattern as Hoops/Pitch/Gridiron.
Personal Graphify — SOTA for Agents
Knowledge Graphs for AI coding assistants — light public map (250 nodes / 179 edges, ~75KB) covering Ava + Scout + Vector. Full private graph for agents; SOTA task compiler + impact + MCP.
Family Neural Architecture
A live Human Design chart tool, not just a write-up — full Type/Authority/Profile, an interactive bodygraph, relationship compatibility, all computed client-side with orbital math verified against real astronomical events.
Contact
Describe the problem — people analytics, workforce strategy, MLOps, or model-building — and I'll tell you if it's a fit.
Select projects · NDA ok · Austin or remote