Cory Policht
FILE · 006 / ABOUT

About.

I build AI platforms, agent toolchains, and spatial systems — end to end. Senior Consultant at Oteemo. Currently building the AI Dev Platform, its agent fleet, and a Quest 3 client that puts it all in space.

Senior Consultant at Oteemo by day. The rest of the year is the platform, the fleet, and the headset: an AI Dev Platform built end to end — orchestration services, sandboxed execution, agent catalogs, and the observability to keep it honest — plus a spatial client on Quest 3 to see the whole system in the room with you.

The decade before was the apprenticeship: SRE and MLOps at Launch Consulting (incident recovery, cloud re-architecture, standing up MLOps for the org), platform modernization at Kalkomey, an independent consulting practice at Sphinx Code, claims and refining systems at Optum and Marathon Petroleum — including an API pass that cut call times fivefold across 17 refineries — and CI/CD and test platforms at USAA, where the functional-testing framework cut end-to-end test time by half and the MISMO standards workgroup turned into two of them. US Army veteran.

VITALS
NameCory Policht
RoleSenior Software Engineer
BasedSan Antonio, TX
NowBuilding the AI Dev Platform
Email[email protected]
StatusBooking selectively for Q4 2026.

TIMELINE ―――――――――

Feb 2025 – Present
Senior Consultant
Oteemo Inc.
2024 – 2026
M.S. Information Technology (in progress)
Western Governors University
Aug 2022 – Aug 2024
Senior Software Engineer & SRE — cloud re-architecture, MLOps adoption, incident recovery
Launch Consulting Group
Apr 2021 – Jul 2022
Senior Software Engineer — Ruby/React platform modernization
Kalkomey Enterprises
Sep 2019 – Jan 2022
Owner — independent consulting: enterprise apps, cloud solutions
Sphinx Code Consulting
Jun 2016 – Aug 2019
Software Developer — CI/CD pipelines, test platforms; MISMO standards workgroups
USAA
2005 – 2013
US Army — veteran
Principles
  1. §01
    Ports, not platforms
    Choose seams over vendors. Domain logic that compiles and tests without its engine survives migrations — Unity to CI, Ruby to React, one cloud to the next. Multi-cloud at Sphinx Code and Launch; engine-free cores at home.
  2. §02
    Fail closed
    On paths that touch hardware, secrets, or production, the default is no. SRE recovery inside 24 hours at Launch, production-support ownership at Kalkomey, and gated model loads at home all start from the same place: systems that can fail must fail safely.
  3. §03
    Evidence-anchored claims
    A fact without a source is a rumor. Measure it — Gatling load tests at Optum, MISMO data-model discipline at USAA, provenance fields in the spatial platform — or cite it: path, line, commit, date.
  4. §04
    Prove it mechanically
    If a boundary matters, the build enforces it. CI/CD pipelines at USAA that halved end-to-end test time, automated functional suites everywhere since, and CI that refuses to compile an engine into the core. Humans review; machines verify.
Stack
AI & Agents
LLM brokering, MCP, agent orchestration, model routing — LangChain, Semantic Kernel, Autogen
Languages
TypeScript, C#/.NET, Java/Spring, Go, Ruby, Scala
Web & Clients
React, Vue, Astro, GraphQL, .NET Minimal APIs
Cloud & Delivery
AWS, Azure, GCP, Docker, Kubernetes/OpenShift, GitLab CI, Gitea Actions
Reliability
SRE & incident recovery, load testing (Gatling), automated functional testing, production support
XR
Unity 6, engine-free C# cores, Meta XR SDK — Quest 3 passthrough MR
READY TO WORK TOGETHER?
Booking selectively for Q4 2026.
Get in touch →