About

I build software that can prove it works.

I'm Eric Jokl, a product engineer who builds AI systems. I take a problem before anyone has fully defined it and carry it the whole way: architecture, code, tests, deployment, and the part after launch where you find out if it actually worked.

The thread through all of it is verification. Most software doesn't fail loudly. A patch looks right and compiles. A cache returns something plausible. A form keeps submitting and quietly stops saving. So I make correctness something a command can answer: a test that fails for a specific reason, a gate that blocks a bad merge, a number measured on the running system instead of quoted.

I've also done the design, the copy, the SEO and the launch myself, then watched real customers use it. That's why I'm quick at the call most handoffs get wrong: which technical decision the business outcome actually depends on, and which one is just preference.

Focus
AI systems, product engineering, and the verification that holds them together.
Building now
MORPH, software that learns a team's decisions from watching them work.
Open to
Full-time engineering roles, contract builds, and early-stage partnerships.
How I learned
By shipping. Formal CS through Harvard's CS50x and Google's professional certificates.
Eric Jokl

Find me

Currently

Building MORPH and taking on product and AI engineering work. Open to full-time roles, contracts and early-stage partnerships.

Start a conversation

Where I'm strong

  • I own the whole path

    Data model, backend, interface, tests, deploy and launch. I've done each of them on real projects as the only engineer, so nothing stalls waiting on a handoff.

    New England Event PlannersCT Party Bus

  • I make correctness checkable

    106 tests on one platform, built around its revenue path. 165 across the systems on this site. Gates that fail the build instead of filing a report.

    MORPHContextGate

  • I build the hard part, not the wrapper

    Program synthesis, automated planning, a semantic cache that refuses to return the wrong answer. The infrastructure around models, not another call to one.

    MORPHPresence

  • I understand why the business wants it

    I've run live commercial sites where search rank and conversion decide revenue. I build with the outcome in mind, not just the ticket.

    CT Party Bus

What I'm still building

  • Most of my work has been as the only engineer. I want the pressure of a strong team: real code review, shared ownership, and people better than me at things I'm still learning.
  • The production systems I've run are small-business scale. I haven't yet operated something with millions of users, and I'm looking for work that gets me there.
  • My computer science is recent and mostly self-driven. I close gaps fast because I learn by building, and the work on this site is the record of that.

What I'm looking for

  • A role on a team building something difficult

    AI infrastructure, developer tools, or product engineering where correctness matters.

  • Contract builds

    A product, platform or automation that has to ship and hold up once it does.

  • MORPH design partners

    A team with one repetitive decision they want turned into tested software.

Start a conversation

How I work

Decide the data model first

Interfaces are cheap to change and schemas are not. I would rather spend an extra day on the shape of the data than a month working around it.

Make failure loud

A system that cannot tell you it is broken will not. Tests that can fail for a specific reason are worth more than coverage numbers.

Write down why

An unexplained decision gets reversed by the next person who sees it. Recording the reason is the cheapest maintenance there is.

Ship, then measure

My opinion about how a page performs is not evidence. Real traffic settles arguments that meetings do not.

Cut what does not compound

I took a Three.js city scene off this homepage. It cost 903 KB and 1.8 seconds of load time to prove nothing. Building it was the easy part; removing it was the decision.

Own the whole path

Design, engineering and business outcome are one problem. Splitting them across three people is where most of the quality is lost.

See the work

Credentials

What I've studied formally.

What each one covered →
  • CS50x: Introduction to Computer ScienceHarvard University · 2026 · verify
  • IT Automation with Python Professional CertificateGoogle · Coursera · 2026 · verify
  • AI Professional CertificateGoogle · Coursera · 2026
  • Data Analytics Professional CertificateGoogle · Coursera · 2026 · verify
  • UX Design Professional CertificateGoogle · Coursera · 2026 · verify