Projects

Things I've built and shipped

Some of this I'm proud of. Some of it taught me what not to do.

# Featured Work

Engineering Emmy Winner

Evercast

Live collaboration for Hollywood

I spent 4+ years building the platform that lets directors and editors collaborate on live video from anywhere. Used on productions like Lost in Space, Chernobyl, and Jimmy Kimmel Live.

  • Scaled WebRTC infrastructure with custom Lua and C extensions
  • Built the live collaboration engine: 4K at around 150ms latency
  • Built the signaling layer, up to 256 concurrent teams
  • Led the team through a pandemic pivot that helped the industry keep working
Current Role

Staff Software Engineer & Engineering Lead

Lumenalta, embedded with a Fortune 100 media company. Data and AI platform.

I lead the data and AI platform behind a Fortune 100 media company, with a team that scales from 8 to 14 engineers as demand shifts. The engineering is the easy part. The rest is walking into a room of VPs and explaining why the AI thing they read about in Forbes won't ship by Q3.

  • I run the data platform: petabytes across BigQuery and Snowflake, and I'm the one paged when a pipeline dies at 2am
  • Own the ML infrastructure, the unglamorous plumbing between a notebook that works on someone's laptop and a model that survives production
  • Turn a year and a half of roadmap into three slides a CFO will actually sign off on
  • Get engineering, product, and finance to agree on a plan that survives contact with reality
  • Mentor senior engineers, and argue architecture in the rooms where the decision actually sticks

# Other Work

## Infrastructure & Scale

  • Global Microservices Migration. Spent a couple of years moving 8+ teams from monolith to microservices. I set the target architecture, wrote the migration playbook, and coached teams through it. Deployment went from weekly to many times a day.
  • Global Database Architecture. Designed the sharding and replication for a worldwide user base. Held query latency under 50ms across regions.
  • Payments and Reliability. Architected a payment system across multiple providers for transactions that could not fail. 99.99% uptime, with automatic failover and reconciliation.

## Data & MLOps

  • Petabyte ETL Platform. Built and ran data pipelines moving petabytes from Databricks, BigQuery, Snowflake, and more. Gave business teams self service analytics.
  • MLOps for Data Science. Built pipelines that let data scientists train, evaluate, and deploy models to production without waiting on DevOps. Model deployment went from weeks to hours.

## AI & Conversational Systems

  • AI Support Agent for Payments and Subscriptions. An AI support agent that answers complex merchant questions from real customer data: transactions, billing history, subscription states. Accuracy is critical and hallucinations are not an option.
  • RAG Pipeline Architecture. Designed the RAG system behind it. It pulls context from a few sources, chunks it semantically, runs hybrid search (vector and keyword), and grounds every answer hard. If it can't cite the source, it doesn't answer.