AI Experience and Product Engineer
I put data to work. I build the experience people use, the AI behind it, and the managed knowledge both draw on, so people, and now agents, get real answers instead of digging for them. I started on the experience side and it's still what I care about most, but I cross every boundary to get there: design, frontend, backend, data, and AI. I want it to be structurally sound and good to use.
My Work
-
JupiterOne Project Tau Rebuilt how JupiterOne builds its AI. There were a dozen AI features, and each one shipped with its own route, agent, tools, UI, and trace. Treated AI quality as a context-building problem and judged it on four questions: correctness, relevancy, intent, and learning. Shipped one agent loop that can stop and ask the user instead of guessing, and one tool bag shared by the in-app agent and MCP servers. The agent uses the UI as its rendering engine. The app brokers every artifact the agent creates, previews, and mutates, so the user decides what lands. Tool calls are grounded in observed facts, which feed a knowledge layer with decay. That layer combines graph, account, and user knowledge, so what the system knows stays fresh and each answer builds on the last one. The product moves from pull, where users explore the graph to dig out answers, to on-demand push, where it tells them what matters and how it connects.
-
Historical Snapshotter Gave the product a memory so it can answer a question security leaders ask all the time: are we getting better or worse? The graph only knew the present, but three platform machines had been writing daily J1QL result rows to S3 for up to 8 years with nothing to read them. Designed and built the pipeline: capture through the existing query-service machinery, transcode to Parquet with DuckDB on EKS, rollup aggregation, and a report reader that serves 32 query templates. That made fast 90-day trends and executive dashboards on historical graph data possible for the first time. It's also the quantitative lane of Project Tau's knowledge. Agents call it to say what changed, not just what is.
-
Experience Rebuild Planned a new web-first client. The plan is one artifact built on one horizontal library of forms and endpoints, with room for an AI companion next to every surface. Designed the migration from 36 independent micro-frontends to a single Vite monorepo with TanStack Router, outlet-composed routing, and a shared primitive layer. Shipped navigation that keeps the URL as the source of truth, plus a universal inspector you can open from every surface in the product, so any traversal can be shared as a link. The migration is done and has been live since April 2026. View release notes
-
Schema Explorer Visual J1QL query builder for people who don't know the schema. Users browse their account's entity-class taxonomy as a focused subgraph, click classes to build traversal chains, and see the J1QL generated as they go. The ordered path is the single source of truth. Chain state, filters, and selection live in the URL, so a half-built query can be shared and survives a refresh.
-
Graph Canvas In-house replacement for a commercial graph visualization library (an estimated $75k/year in licensing), built with React Flow and elkjs. Every behavior was designed so the user can get back to where they were: entity-class packing for spatial memory, node dimming to keep context, and a push-pop inspector stack so traversal never strands the user. No states that need a modal escape or end at a dead-end back button. View release notes
-
AI Platform Started from a blank file in October 2025. SSE streaming with real-time token rendering, a J1QL generation agent, an LLM-as-judge eval pipeline, Bedrock multi-region routing, and a vector embedding cache for natural language querying. The eval harness set the quality bar every later AI feature ships against.
-
SearchAnywhere Built over three major versions across two years, each shaped by what the last one showed about how security engineers search. First a single query entry point, then persistent history and a resizable preview drawer, then full AI integration with multiple query candidates. Announced GA org-wide in March 2025. View blog post
-
Design System Sole owner for three years. Token architecture, MUI 5 migration across 30+ repos, component API design with composable slots, WCAG 2.1 AA compliance, and FedRAMP eligibility work, which opened up government contracts. Replaced a $5k/month hosted visual-testing bill with self-hosted Storybook and Vitest interaction tests.