Me

Sean Koval

seanmkoval@gmail.com · GitHub · LinkedIn · New York City, NY

I use data and AI to tackle complex problems and push toward more general, adaptive systems. In practice, that means building the full ecosystem — models, agents, tools, and evaluation — with a focus on client-facing services in finance and healthcare, where better systems can materially change outcomes.

I love building. Whether it's agents, full-stack applications, data infrastructure, or developer tools — I'm happiest when shipping something end-to-end. My projects span Python, TypeScript, and Rust, from weekend prototypes to production systems.

Sean Koval

What I do

ML Product
Architecture, experimentation, rollout, monitoring. Most of the work happens after the model is already good enough.
LLM Systems
Rubrics, regression suites, human review in the loop. If you can't catch a regression, you don't have a system — you have a demo.
Full-Stack
APIs, frontends, infrastructure. Python and TypeScript mostly; Rust when latency is the requirement.
Data Platforms
Retrieval, semantic registries, dashboards, automation. The goal is a question answered in seconds instead of a ticket filed.

Experience

Senior Data Scientist — Symphony Communication Services Dec 2023 — Present

Led a team of 2, architected RAG + agent features, built LLMOps evaluation pipelines, and shipped full-stack AI products.

Data Scientist — Symphony Communication Services Apr 2021 — Dec 2023

Built real-time topic detection, analytics dashboards, and audio classification pipelines.

Data Scientist — Capital Prawn, Inc. Jul 2019 — Apr 2021

Built quant trading tools, data pipelines in Rust, and crypto market research systems.

Latest writing

View all →

Private Equity's AI Problem Is Not a Model Problem

Why LLM rollouts at investment firms stall out at search, what the four workflows actually demand, and the products that should exist but don't.

GitHub Pulse: Week of March 1, 2026

Agents learned to work in teams, SpacetimeDB collapsed the stack, and WiFi replaced cameras for human sensing.