Custom AI when you need it shipped.
Fractional CTO when you need a partner.
Project-based when you need it built. Retainer-based when you need someone in the room. One operator, nine years of ML / AI — for enterprises and startups alike, from a 2-hour script to a 2,000-hour platform.
Full-stack capabilities. One operator.
If the AI work is real, hire someone with a real ML and data science background — not someone who learned the term last year. That same hire also handles the dashboards, the integrations, the automation, and the call with the team. Whole stack, one hire.
Machine Learning
Real models for real problems — not LLM wrapper theater.
- ›Forecasting, recommendation, lead scoring
- ›Bayesian regression, gradient boosting, deep learning
- ›Productionized with monitoring, retraining, MLOps
Automation
Custom orchestration when it matters. No-code when it's enough.
- ›Custom Python pipelines · Airflow · Cloud Build · GitHub Actions
- ›n8n / Zapier when the problem genuinely fits
- ›The hard ones — granular email-click tracking, dedup logic, rolling analytics
Web & App Dev
Architected to last. Not just shipped to demo.
- ›Architecture from scratch — framework, DB, deploy topology
- ›Multi-tenant SaaS, B2B2C, internal platforms
- ›Built to keep working in 5 years — not 5 weeks
Integration
Any system. First try. Still working in five years.
- ›CRMs, clouds, AI/LLMs, social, payments — agnostic
- ›Event-driven: webhooks, queues, idempotent retries
- ›Debuggable when the upstream API changes (and it will)
AI Ad Studio
$10K studio ads for $500. In a couple hours.
- ›Empty venue → photographed scene → director-quality video
- ›Real composition, not generic generative slop
- ›Per-spot cost a fraction of an agency
Fractional CTO
An engineer who can also talk to your board.
- ›Architecture review, hiring, vendor selection
- ›Roadmap → execution → measurement
- ›Comfortable with founders, ICs, and Fortune 500s
Real systems. Real revenue.
Avocado Demand Forecasting
Built a Bayesian regression model that beat the open market for 3 consecutive years for one of the nation's largest avocado suppliers.
$300M Bidding Engine
Designed and deployed a large-scale product recommendation and bidding optimization engine for an auction company processing $300M+ annually.
Storywarz
Comedic deceptive-storytelling party game. Real-time rooms, no-account play, truth-or-bluff voting, and a scoreboard that rewards a good liar.
Song Selfie
Make AI songs about yourself. Venue partnerships, automated revenue splits, share-ready audio — built end to end.
Three things that make a difference.
Same person, every layer.
Model, app, pipeline, integration, dashboard, board update — same operator on each. Hire for any one of them and the rest comes with.
Real code where it matters.
n8n and Zapier when the problem genuinely fits. Custom Python, Airflow, Cloud Build, GitHub Actions when it doesn't — and that's most of the time.
Still in production, years later.
One predictive model from this stable has been running in production for 9 years — same algorithm, same coefficients, same client. The work is built to outlast the engagement.
A client wanted to track every individual click across their email campaigns — with deduplication, attribution logic, and a rolling analytics layer. That's not a Zapier problem. It's a custom Python pipeline, in GitHub, running on Cloud Build, with idempotent logic that doesn't double-count.
Most engagements look like this — somewhere on the path from “we'll glue these tools together” to “wait, this needs real engineering.” That's exactly where the right hire matters.
If you can describe it, we can ship it.
Enterprise teams need someone who can talk to the board and write the model. Startups need someone who can choose the stack and ship by Friday. The right operator does both — that's the whole point.