Work

AI products, from research to production.

The common thread is taking AI capabilities and making them useful: video generation, agentic workflows, code review, retrieval systems, optimization, and product quality.

Start a Conversation Download Resume

Current

Founding Engineer, Cubic

AI product engineering for code review

  • Leading AI development for a code review product publicly ranked #1 on an independent benchmark.
  • Working across model behavior, codebase context, evaluation, cost, product quality, and developer experience.
  • Turning review signal into a product engineers can trust inside real repositories.

2025

Founding AI Engineer, Nexcade

Freight quotation workflows with LLMs and agents

  • Built multi-agent quotation automation for real operational workflows.
  • Designed systems around workflow orchestration, validation, supplier data, and human oversight.
  • Helped move the product from early AI idea to usable workflow infrastructure.

2024 - 2025

Tech Lead and Technical Advisor, KIRO

RAG chatbot infrastructure and AI product execution

  • Led development of production RAG and agentic workflows for a financial coaching product.
  • Worked across retrieval, product constraints, interface quality, and fundraising-stage execution.
  • Supported the transition from prototype to real product usage.

2021 - 2023

Principal AI Engineer, Rephrase AI

Generative video systems, acquired by Adobe

  • Led core text-to-video and lip-sync technology work for personalized video generation.
  • Managed research and engineering direction across model quality, production constraints, and cost.
  • Contributed to technology later acquired by Adobe and integrated into its generative AI ecosystem.

2015 - 2021

Deep Learning Research Engineer, Samsung R&D

Optimization, computer vision, and production ML systems

  • Worked across deep learning optimization, cost reduction, latency, and deployment constraints.
  • Partnered with multiple internal teams to translate research ideas into usable engineering systems.
  • Built the technical foundation that later carried into startup and product work.

Pattern

How the work usually shows up.

From capability to product

Model behavior only matters after the workflow, interface, cost, and failure modes make sense.

Quality and constraints

Evaluation, latency, reliability, and cost are product decisions, not cleanup work after the model is chosen.

Technical direction

The useful decisions are often what to measure, what to ignore, and what the team should stop doing.

Start a conversation

For talks, collaborations, advisory conversations, or selective senior opportunities, send a short note with the audience and context.

Start a Conversation