LocationDallas, TX · GMT-6
RoleSr. Full-Stack Eng. · AI Agent Specialist
StatusOpen to AI consulting + collaboration

I build AI-native frameworks
that build shipping products./01

Skills, agents, hooks, and instructions composed into a system small teams can lean on. The result: production-grade software in less time, with the rigor of a senior engineering review baked in.

10+ yrs
Production engineering
135+
DevRunway tech layers
5
Domains shipped

A decade in production. Now teaching software to think.

Engineer first, AI specialist second. The thing that hasn't changed: shipping software that enterprise teams actually use, in domains where mistakes cost money.

I'm a senior full-stack engineer at IntelliSavvy in Dallas, where I built PolicyAI — an underwriting agent that turns multi-day policy reviews into conversations. Outside work, I build open-source developer tooling: DevRunway turns Claude Code into a stack-aware engineering teammate with 135+ technology layers, 35 quality hooks, and a full SDLC workflow from first idea to production deploy.

The interesting work isn't getting a model to answer — it's wiring it into a system where someone's job depends on the answer being right.

Before AI agents I shipped insurance broker portals, healthcare platforms serving thousands of daily users, hybrid mobile apps, and financial visualizations. Across React, Node, AWS, and a long tail of whatever the problem required. The constants are care about the people using it and care about the people maintaining it next year.

Right now I'm interested in agent design patterns that hold up under enterprise constraints — auditability, deterministic fallbacks, structured tool use, and what "good UX" looks like when half the interface is non-deterministic.

Specialized in agents. Built on a full stack.

Eight categories of tools I reach for. The first one is where most of the new work lives; the rest are how it ships.

01 / Specialty
AI Agent Development
Building intelligent agents that fit inside real enterprise workflows. LLM orchestration, tool use, retrieval, evals.
LLM Integration Conversational AI NLP Prompt Engineering Intelligent Automation AI-Powered Apps
02 / Frontend
Web Technologies
UI systems and component architecture for dense, data-heavy product surfaces.
ReactTypeScript ViteMaterial UI D3Redux MobXSass
03 / Backend
Server & APIs
Microservices, API design, integrations. Mostly Node, with .NET / C# where the domain calls for it.
Node.jsExpress Sails.NET / C# Microservices
04 / Cloud
AWS
Production infra. The boring, important parts of putting agents into customer hands.
EC2Lambda RDSCognito Step FunctionsAmplify Route53IAM
05 / Delivery
CI/CD & Version Control
Pipelines that get changes to production with a paper trail.
TeamCityJenkins DockerGit GitLabBitbucket
06 / Mobile
Native & Hybrid
React Native and Ionic apps shipped through both stores. Three thousand+ downloads on the healthcare app.
React NativeIonic CordovaApp Store Play Store
07 / Data
Databases
Relational and document. Schema design through query optimization through the migration that should have been simple.
PostgreSQLMySQL MS SQLMongoDB NoSQL
08 / Quality
Testing & Ops
Unit tests, code coverage, observability, security scanning. The stuff that makes agents trustworthy.
RTLJasmine DatadogSonarQube CheckmarxAuth0

The flagship, and the systems behind it.

Flagship · Open Source · MIT · 2025 — present

DevRunway — stack-aware SDLC for Claude Code.

Turns Claude Code into a stack-aware engineering teammate for any technology stack. Run /setup once — answer 35 questions about your frontend, backend, database, auth, cloud, CI, and tooling — and DevRunway generates stack.json + .mcp.json and installs only the standards that match what you're actually building. Every command, hook, and quality gate adapts automatically as you touch different files.

135+
Tech layers
35+
SDLC commands
93
Eval test files
Claude Code MCP SDLC Automation Sub-agents Open Source MIT
Visit devrunway.dev GitHub ↗
/SETUP DevRunway 35 q · 6 screens YOUR STACK React · Node · AWS · Postgres stack.json .mcp.json GitHub · Jira · Figma LAYERS 135+ tech modules 35 HOOKS secrets · lint · security FULL SDLC brainstorm → prod
Live & open source
Past work

A timeline, roughly chronological.

Five roles, three countries, one consistent thread: software people rely on, shipped.

2021 — now

Sr. Full-Stack Developer & AI Agent Specialist

IntelliSavvy · Dallas, TX
  • Pioneered PolicyAI, the AI agent that transformed how underwriters interact with applications — multi-day reviews into conversations.
  • Built the broker portal that cut process time 70%; Node APIs and React surfaces that improved issuance throughput 60%.
  • Hit 99% CSAT by replacing legacy workflows with the latest of the web platform.
  • Reduced bug rate 25% through code review discipline; optimized API latency 25% using Datadog.
  • Owned deployment across 5 AWS environments.
2021

Senior Software Engineer

Dun & Bradstreet · Hyderabad, IN
  • Architected secure React + .NET Core stack; delivered 50 backend APIs.
  • Led training sessions that lifted team efficiency 40%.
  • Cut CI/CD deploy time 60%; resolved 50 code-security issues.
2017 — 2021

Tech Lead

ValueMomentum Software Services · Hyderabad, IN
  • Designed 30+ UI screens in Material UI / React / Angular.
  • Built ~100 Node microservices across the product line.
  • Implemented SSO across 5 platforms; managed AWS infra at 99.9% uptime.
  • Reduced bug rate 45%; lifted team efficiency 25% on agile delivery.
2015 — 2017

Technology Analyst

Infosys Ltd · Hyderabad, IN
  • Translated client needs into 200 user stories; shipped 3 web apps in Angular / React.
  • Designed 50 responsive screens; implemented localization, keyboard navigation, screen-reader support.
  • Cut defect rate 35% through review & test discipline.
2013 — 2015

Senior Software Engineer

Icrea (Suja) Infotech · India
  • Launched a healthcare web app for 25 providers / 500 daily users.
  • Built a hybrid mobile app (3K+ downloads) and 50-table backing schema.
  • Deployed to Windows servers supporting 4K users; reduced bugs 25% with unit-test rollout.

Building an agent, or thinking about it?
Let's talk.