By day, I'm a Senior Engineer at AWS building production generative AI applications.
Outside work, I create open-source tools and write about the hard problems most teams skip when shipping enterprise AI.
FootPrint — The flowchart pattern for backend code
Business logic becomes a directed graph that produces causal traces an LLM can reason over.
- 7 flow patterns · transactional state · PII redaction · auto-generated tool descriptions
- 6 modular libraries:
memory·builder·scope·engine·runner·contract - Parallel fork/join · streaming · patch-based state · time-travel replay
npm install footprintjs
Enterprise Gen AI Application — LinkedIn newsletter (320+ subscribers)
| # | Post | Core idea |
|---|---|---|
| 1 | From Supply-Driven to Demand-Driven | The chatbot should drive UX, not assist it |
| 2 | Make Search the First Tool | STAY/SWITCH + Focus Token protocol |
| 3 | The Flowchart Pattern | Making backend code self-explainable for AI |
Bridging UI Design and Chatbot Interactions
Applying form-based principles (Submit/Reset → STAY/SWITCH) to conversational agents.
Published at HCI International 2025 · Springer proceedings
Visible Reasoning
A framework for deterministic LLM agent transparency — a "third paradigm" distinct from chain-of-thought and LLM-as-judge.
Accepted at HCII 2026 · Springer proceedings
Weave (data vis sessions)
→ StateTree (state diffing)
→ FootPrint (execution graphs + causal traces)
→ AgentFootPrints (LLM adapters)
10+ years on one problem: making the internal state of complex systems legible to whoever needs to understand them — first humans, now AI.
PhD in Computer Science, UMass Lowell · Dallas, TX
LinkedIn · Medium · Google Scholar