About

Background & approach.

Background

Who I am.

I'm an AI-Native Engineer with a Senior SDET background, building agentic AI systems and the evals, guardrails, and quality practices that make LLM products reliable in production. Nine years across healthcare, enterprise SaaS, and developer tooling means I bring deep quality engineering instincts to AI-native product work, things like LangGraph agents, MCP integrations, RAG with pgvector, and human-in-the-loop workflows.

My approach treats AI as a collaborator, not a replacement for judgment. The most useful AI features are the ones that ship, stay observable, and earn user trust, which means owning the whole stack from prompt and context design through retrieval, orchestration, evals, and the product surface where users actually meet the model.

Principles

How I build reliable AI.

Reliable AI

Guardrails, evals, monitoring, and human-in-the-loop turn LLMs into production systems instead of demos. Quality is the difference between an AI feature that ships and one that gets rolled back.

AI as a collaborator

Agents accelerate humans, they don't replace judgment. The best AI features amplify expert decisions, leave audit trails, and stay legible to the people who own the outcome.

End-to-end ownership

Shipping AI-native features means owning the whole stack, prompt and context design, retrieval and embeddings, orchestration, evals, and the product surface where users actually meet the model.

Cross-functional collaboration

The best systems are built when engineers, product, design, and domain experts are aligned early. Quality comes from shared understanding, not from a final QA gate.

Smart automation

Automation should free teams for the work humans do best, exploratory testing, ambiguous edge cases, and product judgment, while critical paths stay covered without manual effort.

Continuous improvement

AI systems drift, models change, and prompts decay. Tight feedback loops, regression evals, and observability keep features honest as the underlying technology moves underneath them.

Stack

Tools I reach for.

Testing

PlaywrightSeleniumCypressRobot FrameworkPyTest

Languages

TypeScriptJavaScriptPythonC#DartKotlin

Frontend

ReactAngularNext.jsAstroTailwind CSS

Backend

Node.jsDjangoFastAPI

Cloud

AWSGoogle CloudAzureFirebase

Tools

GitDockerJiraCI/CD

Let's connect.

Always interested in discussing agentic AI systems, LLM evals and guardrails, AI-native product engineering, or potential opportunities.