Forward-deployed AI engineering · Austin, TX

Stuck software, shipped - for a fixed price.

Two kinds of work: legacy products that can't be modernized the old way, and AI systems that never made it out of the pilot. Either way it ends up running in production, for a price agreed before we start.

The first conversation is free. Then a short paid scoping, credited in full toward the build.

Modernize a legacy product →Finish or build an AI system →

20 yearsbuilding software and firmware
3× CTOco-founder of venture-backed companies
M&A and IPOled teams from founding through exit
100M+monthly users on products I've built
FDA 510(k)software lead on a ring cleared for sleep and blood pressure
Built with and for ZyngaRocket GamesDrive.fmLinusHappy HealthPalo Alto Strategy Group

What you can buy

Start with scoping. Then one fixed price for the result.

You know the price before the work starts. If it takes us longer, that's ours to absorb.

Start here

Scoping

One week for an AI system, two for a legacy product. You get the outcome, the design, the acceptance criteria and a fixed quote, in writing. The fee is credited in full toward whichever of the three you choose, and the document is yours either way.

1–2 weekscredited in full toward the build
Fixed price · up to 6 weeks

Production Sprint

One thing, finished: a stalled AI system taken to production, or the first slice of a legacy product rebuilt. It has to fit in six weeks.

  • Running in your environment, under your identity and permissions
  • Verified against the acceptance criteria
  • Monitoring, deployment and a clean handoff
Fixed price · delivered in stages

Build

Everything that doesn't fit in six weeks: a new product from nothing, or the full rebuild of a legacy one.

  • Designed, built and verified end to end
  • Integrated with your data, tools and approvals
  • Accepted against criteria we agree up front
Flat monthly fee

Engine

A Product Engine that owns a product line after launch, with one senior owner answering for it.

  • Ships, monitors, fixes and improves continuously
  • Works in your Slack or Teams, GitHub, Jira or Linear
  • One accountable owner for decisions and results

Scoping tells you which one you need. If the result fits in six weeks, it's a Sprint. If it's bigger, it's a Build, and its first stage is usually a Sprint, so you see something running before you commit to the rest.

Selected work

Private equity, healthcare, consumer.

Different industries, the same job: make it run where the real constraints are.

Palo Alto Strategy Group
Private equity · technology due diligence
Prototype to production

A single-file prototype rebuilt as the firm's production diligence platform.

PASG runs technology due diligence for private-equity buyers. They had a single-file prototype. We rebuilt it as the production platform the firm uses to perform due diligence in a consistent, structured and benchmarked way: it codifies the firm's models and benchmarks, checks reports for consistency and quality, and validates findings. The platform is the subject of a pending US patent, with Cory as a named inventor.

BenchmarkingConsistency checksFindings validationPatent pending
Happy Health
Healthcare · sleep medicine
Build

Software leadership behind an FDA-cleared ring, then the clinical systems around it.

Happy Health's ring is FDA-cleared for sleep and, as of 2026, holds the first FDA clearance for cuffless, calibration-free blood pressure. As VP Software, Cory led the software side of the 510(k) work. Since then we have built the clinical charting system its providers work in, live since October 2024, and wired it to claims and payments.

FDA 510(k)Sleep and blood pressureClinical workflowsClaims and payments
YourRealm
Consumer · aging independently
Engine

From nothing to a live app, with one principal.

YourRealm helps older adults stay independent. It started with no company, no repository and no engineering team. One principal and a Product Engine took it from first conversation to a live app.

GreenfieldiOSProduct Engines

How we work

I run an AI engineering team the way I ran human ones for twenty years. I set the architecture, review what matters and approve every release. The team writes, tests and monitors around the clock, and every decision is on the record. You get a team's output with one senior owner answering for it.
Cory Johnson

Inside your systems

We work in your Slack or Teams, your repositories and your ticketing, under access you control. Nothing leaves that you haven't approved.

One senior owner

The person who scopes the work is the person who answers for it. No account managers, no handoffs to a junior team.

Verified, on the record

Every release is tested independently before it ships, every consequential change has a human approval, and you get a readable record of what was decided and why.

About Cory

Twenty years of shipping. Three of them on nothing but AI in production.

I've built software and firmware from game consoles and wearables to cloud systems serving hundreds of millions of people. I've co-founded three venture-backed companies as CTO, led engineering through an acquisition, and taken medical device software through FDA clearance.

For the last three years I've done one thing: put AI into production inside real businesses. ilucara is how I do that for clients.

  • Founderilucara · Product Engines
  • Co-founder & CTOLinus · Drive.fm · Sol
  • VP SoftwareHappy Health
  • Studio CTOZynga
  • Director of Eng.Rocket Games
  • EducationGeorgia Tech, Computational Media
  • Patent filingsReal-time audio (2018) · due diligence platform (pending)

Get in touch

Tell me what's stuck.

Thirty minutes, with me. Bring the system, the constraint and the deadline. You'll leave knowing whether we can ship it and what it would take.