Thirteen years inside the pipelines that carry a performance from one language into another. Five more putting generative models into those same pipelines, for the studios and broadcasters who own the content.
Thirty-two productions across dubbing, subtitling, scripting and audio description, 2016 to 2025.
Before I sold any of this, I ran it. Dub stages, scripting, timing, talent, rights, delivery. The unglamorous middle of the process, where localization actually lives or dies.
Ran Amazon Prime's English dubbing account end to end: casting and talent, studio scheduling, rights, delivery. Forecast and bid against $80M+ in production budget, and was the standing commercial contact for Amazon, Disney and Netflix. Built the automated invoicing and payment system the dubbing talent were paid through.
Built internal tooling that automated scripting inputs, box timing and shot detection, cutting subtitle and dub turnaround by 80%. Led the rollout across production, which meant teaching the pipeline to people who had done it by hand for years.
Lead editor for Amazon Prime, including every Hindi original. Rewrote the company's proprietary localization software so it would render Japanese and Hindi correctly. Wrote audio description for Disney, Netflix, Amazon, Warner Bros. and NBCUniversal titles.
Not demos. Real deployments, in front of real rights holders, with the legal and talent questions answered well enough that the work shipped.
I take an AI dubbing and localization platform to market with studios, agencies and broadcasters. It is listed on AWS Marketplace and was demonstrated on the AWS Elemental stand at IBC 2026. The platform is built on ElevenLabs voice models, so I have spent the past year putting them in front of rights holders and answering the consent, talent and legal questions that decide whether a deployment ships or stalls. Separately, I originated the company's media data business: consent-chain speech sourced through the localization providers who hold the talent agreements, which our CEO adopted as the strategy for the vertical.
First commercial hire in North America for a real-time AI dubbing product across 140+ languages. Designed and ran the custom localization builds for live sport and global events, and landed the first paid pilots (WWE, the Kansas City Chiefs, MLS) from cold outreach, then sat with each of them through the work.
What I make when nobody asks me to. All of it shipped; none of it by a developer.
A Michigan message board that had run continuously since 1995 lost its host and died. I rebuilt it in six days by directing an AI coding agent. Python, SQLite, plain HTML, $7 a month. I kept the 1998 look the members wanted and added photo uploads, polls, live scoreboards and two custom games. It has been live and busy ever since with its ~350 members. I am not a developer. Two years ago this would not have been possible for me.
The write-up includes the rebuild timeline and the incident log: the real production failures and what I did about each one.
A crime story travelling from the American Midwest to Singapore. The formal conceit inverts the two worlds: the American half shot steady and fully lit, the Singapore half captured rogue on iPhones. No pre-written dialogue. A precise shot list and structure, with actors improvising inside hard constraints. Built as a 40-card dialogue-free spine.
The part that came first, and the reason any of the rest of it reads as craft rather than throughput.
Co-host of a weekly Michigan football show, second season. Recorded and published on a fixed weekly cadence. An hour of talk that has to be shaped, not just captured.
Read and covered features for acquisition; ten of the scripts I championed were greenlit, together taking roughly $300M at the box office.
Written at Tisch's Singapore campus, where I lived for three years. Feature scripts include Motor Lion City and Cineribus. Earlier work archived at nccoston.com.
Every job I have had sits on the same seam: a performance made once, in one language, that somebody now needs to exist everywhere, in everyone's voice.