Technical Project Manager · Delivery, Release & Launch · Live-service and platform delivery
Cross-team delivery, release operations, applied AI.
Seven years at AccelByte, from the docs to running release and launch for live-service games. Remote worldwide (contract or EoR) or employer-supported relocation.
- 8+ live-service launches in the past year
- Delivery across more than five service teams
- One major service release a month, plus near-daily hotfixes
- Incident commander in the launch war room
What I'm hired for
A release date is a promise a whole company has already made. My job is to make it survivable: readiness review that catches what the plan missed, a change gate that actually holds, dependency tracking that surfaces the collision before the date does, and someone to run the room when a launch window degrades.
- Release and launch owner. You ship on a live cadence and need someone to own readiness and the war room, not just the tracker.
- Cross-team delivery lead. Your teams each own their own services, and the dependencies only become visible when they bite.
- Delivery systems inside product operations. You are scaling and need the operating system built: intake, engineer-owned estimates, change control, post-launch reporting.
What I'm not. A creative producer with shipped title credit. A domain specialist in credit risk, brokerage compliance, healthcare, or localisation. An engineer. I have worked next to all of those people as the delivery owner, and I would rather say so here than have you find out in round two.
What I actually did
- 1 Built the change gate every production change had to pass, with impact, QA status and a named approver.
- 2 Ran the war room as incident commander when launches and releases broke.
- 3 Built the Grafana launch dashboards the team checked before each launch.
- 4 Owned delivery for a portfolio of game studios at once, indie to AAA, for over two years.
How I got here
Hover or tap a stepBuilt the change gate, the launch-readiness checklist and the Grafana launch dashboards. Incident commander in the war room.
Start with these
All 15 case studies →What I'm building now
Since leaving AccelByte in July 2026 I've been learning the modern stack the hands-on way: infrastructure, ads and analytics, LLMs in production. I direct the AI, then check its work.
- An AI assistant for small merchants live
- A commerce storefront live
- A payment aggregator on a licensed provider live
- A catalog site with search analytics live
- A self-hosted commerce platform migrating
- An internal company OS in build
- And off-screen: 3D design and CNC machining, from model to toolpaths to G-code to a cut part, testing where AI actually helps.
What this means outside games
Platform and infrastructure delivery, in the language of the hiring team
The work is the same; the vocabulary is not. If you run infrastructure, enterprise delivery or data-centre operations, this is the mapping. Every line below is something I already owned — none of it is a pivot I am promising to make.
- Release train and required change gate — every production change carried impact, downtime, QA status, a named approver and one thread of record. Maps to: production change governance and auditability for SLA-bound services.
- Launch readiness across ~25 customer accounts and six owner teams — with no direct authority over any of them. Maps to: cross-team delivery governance in matrix organisations.
- Launch-health dashboards and post-launch reporting — built in Grafana, reviewed by leadership to make release decisions. Maps to: observability and executive reporting for capacity and reliability.
- Customer blockers turned into workarounds and feature proposals — options pitched with trade-offs, then the accepted one taken through prioritisation and delivery. Maps to: solution ideation with clients, not ticket-taking.
- Incident command through launch windows — war room, resolution, then RCA rather than blame. Maps to: incident and problem management discipline for on-call organisations.
- Six AI systems in production — spend ceilings, deterministic validation between the model and anything it changes, approval boundaries on irreversible actions. Maps to: AI operations guardrails before an enterprise deployment, not after.