Biography

I work at FTAI Aviation in New York, where I build the retrieval and agent infrastructure behind customer-facing AI features. That means a RAG backend in Python and FastAPI, multi-tenant agentic workflows exposed over Model Context Protocol servers, and the Azure DevOps pipelines that get all of it onto Kubernetes without drama.

Before that I spent two years in computer vision research at Washington University in St. Louis, across the Multimodal AI Lab and the Digital Innovation Accelerator, with a parallel stint at SLU’s Multimodal AI Lab. I have contributed to work on satellite image synthesis and cross-view pose estimation, and I led computer vision for WashU’s Ursaworks robotics team, which is where I learned what a latency budget feels like when it is set by physics rather than by an SLO.

The thread running through all of it is the same problem from different angles: research code demonstrates that something is possible, and then somebody has to make it dependable. I have mostly been that second person, and I have come to like it.

Skills

Emphasised entries are the ones I use in production most weeks. There are no proficiency bars here, because a bar implies a measurement nobody took.

Languages

Python and SQL daily at FTAI. Rust and C++ on systems and infrastructure work.

  • Python
  • SQL
  • TypeScript / JavaScript
  • Rust
  • C++
  • Node.js
  • HTML / CSS

Backend and APIs

The service layer for every AI system I have shipped.

  • FastAPI
  • REST APIs
  • OAuth2
  • Server-sent events
  • React
  • Next.js
  • Gradio

AI and machine learning

Production retrieval and agent systems at FTAI; model training in research settings.

  • RAG
  • LangChain
  • Agentic systems
  • Model Context Protocol
  • PyTorch
  • Model evaluation and benchmarking
  • ONNX
  • TensorRT
  • FAISS
  • OpenAI / LLM APIs

Data and storage

Postgres and Redis are the pair I reach for most; the rest come out for scale.

  • PostgreSQL
  • Redis
  • Apache Spark
  • Apache Kafka
  • BigQuery

Infrastructure and delivery

Containerised services on Kubernetes, shipped through Azure DevOps pipelines.

  • Linux
  • Docker
  • Kubernetes
  • Azure DevOps
  • CI/CD
  • Observability and logging
  • Terraform
  • GitLab CI
  • AWS
  • Microsoft Azure
  • GCP
  • HPC Slurm

Testing and tooling

Automated coverage is the reason the release pipeline is allowed to be fast.

  • pytest
  • Automated testing
  • Git
  • ruff / mypy
  • Prometheus
  • Postman
  • CMake
  • Bazel

Education

Bachelor of Science, Computer Science & Mathematics

Press

Students in summer program develop AI tools, advance faculty research

Coverage of the Digital Transformation Summer Corps, including the vision-language system my team built for assessing neighbourhood walkability, greenery and safety. I described the combination of street-view imagery and language models as a way to “speak” with a city.

Personal

Outside of work I play board games, travel when I can, and play basketball. I used to volunteer as an EMT for my local town, which is still the most direct reminder I have that competence under pressure matters more than looking busy.