AI & Agents
- Google ADK
- Model Context Protocol (MCP)
- Agent-to-Agent (A2A)
- OpenAI · Anthropic · Gemini
- LLM Orchestration
- Tool / Function calling
- Prompt engineering
- Responsible AI
Senior Software Engineer at Walmart Global Tech with 13+ years across distributed backend services, GraphQL federation, and large-scale test automation. Currently shipping production AI agents on Google ADK with multi-model LLM orchestration and MCP-based tool integration.
I'm a Senior Software Engineer at Walmart Global Tech based in the DC / Reston area, where I build backend services and AI agents that power Walmart's Auto Care Center (ACC) ecosystem. My work spans Node.js and Java services, Apollo GraphQL federation, Cosmos DB, Kafka, and increasingly — agentic AI systems on Google ADK with the MCP protocol.
Earlier in my career I led test automation at scale — building frameworks for React Native apps, deploying Report Portal dashboards, and standing up parallel mobile UI testing on Jenkins. That QA foundation still shapes how I write services: everything I ship is instrumented, contract-tested, and load-validated before production.
Lately I've been focused on the Sparky assistant — an
agentic shopping experience that orchestrates GPT-4o, Gemini 2.5, Claude Sonnet, and
Claude Haiku through a single ADK LlmAgent with tool calls into Walmart
MCP servers, Element Gateway, and remote A2A agents.
A selection of the projects I currently own or have shipped at Walmart Global Tech. Most are internal repositories; descriptions are written so they make sense without access.
Reference implementation of progressive disclosure with Skills for
Walmart's Sparky shopping assistant. A single ADK LlmAgent
orchestrates product search, cart, checkout, and order tracking against Walmart's
MCP servers — with four interchangeable orchestrator models (GPT-4o, Gemini 2.5
Flash, Claude Sonnet 4.6, Claude Haiku 4.5) exposed as separate ADK entry points.
LlmAgent with provider-switchable runtime configurationweb_search and remote A2A recipe_agentGenerative AI agent powering Walmart Auto Care Center experiences. Owns intents for license-plate lookup, parts finder, buying guides, and tire-size resolution. Built from the ground up: rate-limiting configuration, structured logging, code-driven messaging, and an A2A integration for the parts finder skill.
A purpose-built MCP server that automates large-scale, multi-module refactors across ACC microservices. Exposes typed tools to LLM clients so an agent can confidently plan, propose, and apply repository-wide refactors with diff previews and guardrails.
Primary backend orchestration service for Walmart's Auto Care Center. I lead day-to-day releases — pre-release management, CPU and scaling tuning, SonarQube and Snyk gate hardening, Looper-based integration / functional / E2E test orchestration, and TestHub executor configuration. 30+ merged PRs in the last quarter alone.
Spring Boot microservices that resolve vehicle identity across the ACC experience — from VIN, license plate, year-make-model, and Carfax-sourced metadata. Owns the Sparky-Brain application wiring in the service registry and stage-wave deployment config across multiple environments.
End-to-end automation platform for ACC services — functional, contract, and E2E tests wired into Looper executors, with deployment-window gates and TestHub reporting. Built on the React Native automation foundation I led earlier (Cucumber + Spring Boot, parallel iOS / Android execution, simulator-availability routing).
acc-omni-service: CPU / scaling tuning,
SonarQube + Snyk gate hardening, Looper executor configuration, pre-release and
main-branch governance.
I'm always happy to chat about agentic AI, MCP tooling, GraphQL federation, large-scale automation, or career paths from QA into AI engineering.