Zero to exit — nationwide platform scale
Co-founded and led engineering for 13 years. 200K+ users, async tournaments, realtime stats, board seat through acquisition.
200K+ users5K+ schools13yr tenure
Work
Case studies
Deep dives on platform scale, org building, and practical AI — written for founders and engineering leaders.
Experience at
Real-time matchmaking engine — first in scholastic esports
No queue-based matchmaking existed in scholastic esports. Built a game-like real-time system: students form parties, parties are matched against each other live via WebSocket connections, with edge case handling for brittle real-time states. Eliminated the forfeit rate problem that plagued scheduled matches. First system of its kind deployed at nationwide scale.
Real-timeWebSocketsMatchmakingNode.js
Challenges system — stat-based leaderboard competition
Built a stat scraping pipeline to ingest heavy in-game data from publisher APIs (Riot Games and others), apply a ranking algorithm based on player performance, and surface live leaderboards. Built to scale using message queues to pipeline processing and handle failures gracefully. Powered the async tournament model that differentiated the platform.
Message queuesStat pipelinesRanking algorithmScale
Computer vision auto-clipping
Applied visual recognition to live game streams to detect in-game highlight events (goals, key moments) and automatically generate clips for VOD distribution. Built in Python. Predates the current LLM wave — genuine ML engineering on video data when there was no API to call.
Computer visionPythonVODML
Blueprint.study — build vs. buy decision that drove the acquisition
Gen Esports was paying a third-party LMS vendor as part of the sales package. It was eating into margins and didn't meet our quality bar. I took it upon myself to replace it — co-designed it with internal stakeholders and educators, led development with one additional engineer, shipped fast with a startup mentality inside a growing company. $3M+ ARR. When PlayVS acquired Gen Esports, Blueprint was the primary thing they were buying.
$3M+ ARRBuild vs. buyNext.jsLAUSD
AI adoption program at PlayVS — 8 engineers outshipping 30
Inherited a team of 8 engineers at a downsizing company and a mandate to shift from large-org mindset to startup execution. Ran round tables, pair programmed with individuals, created AI tool budgets, built QA skills the team could use, and implemented a documentation practice specifically to give AI tools context about legacy black-box modules. The result: 8 engineers shipping a payment processor, Live Training (Google Meet-style seminar interface), a proprietary student-to-college matching algorithm, Discord integrations, and more — faster than the previous larger team had been.
AI adoptionTeam transformation8-person teamVelocity
Blueprint.study — LMS shipped under deadline pressure
Next.js/Vercel LMS deployed in major districts including LAUSD. $3M+ ARR and a key driver of the PlayVS acquisition.
$3M+ ARRNext.jsLAUSD
Prototyping AI workflows at PlayVS
Rapid prototypes for ops automation and engineering tooling — contributing to measurable gains in operational efficiency and developer velocity.
−60% ops toil+30% velocityPrototypes
AWS → GCP — 66% cost reduction
Full infrastructure migration with Google Cloud engineering support. Improved performance while cutting monthly spend.
−66% costGCPK8s
Want the quick version? Ask the chat assistant on the home page.