Lead Software Engineer · Frontend Architecture · Agentic AI
Cox Automotive · Xtime · Texas, USA

I'm a full-stack engineer with 18+ years of experience building large-scale enterprise and SaaS applications across automotive, banking, retail, and e-commerce domains. At Cox Automotive, I lead frontend architecture for Service Quoting and Repair Order workflows powering franchise dealers across North America, and Canada.

My current focus is agentic AI, Generative AI, and workflow automation — applying Claude Code, Claude Skills, spec-driven development, and RAG pipelines to real production challenges. I mentor engineers, author internal standards, and drive cross-squad AI initiatives that move the needle on measurable engineering productivity.

• LinkedIn: https://www.linkedin.com/in/raja-pasupuleti

• 9+ years at Cox Automotive (Xtime) — Lead Software Engineer on Service Studio, Service Quoting, and Repair Order (RO) cloud applications
• Orion Systems Integrators — Sr. Software Developer; delivered Xtime Service Studio, CitiRisk (Citibank), Compass group and Wells Fargo iReports trading platform
• IBM India — Sr. Application Developer on Walmart Global Replenishment System (GRS) — 8 integrated microfrontend web apps using ExtJS & JSF
• Melstar / Reasoning Global — delivered enterprise solutions for ServiceMaster, MartJack e-commerce, and school communication portals
• Consistent record of proof-of-concept delivery, reusable library authorship, and cross-team technical leadership

• Frontend Architecture — React, TypeScript, Micro-frontends, Module Federation, design token systems, Spec-driven development contracts
• Agentic AI & Gen AI — Claude Code workflows, Claude Skills, CLAUDE.md constraints, 41-prompt reusable library, custom agents, context tiering
• Workflow Automation — AI-assisted spec-driven delivery, cox-spec-kit adoption, GitHub spec-kit integration
• RAG & Knowledge Systems — pgvector, LangGraph, RAGAS evaluation, SharePoint/Rally ingestion pipelines for product knowledge base
• Full-Stack Delivery — Spring, Hibernate, REST, Node.js, Oracle/PL-SQL, AWS
• Engineering Quality — TDD, spec-driven development (EDD), 20-metric code health scoring, ADR authorship
• Mentorship — Active mentor to 4 engineers; cross-squad knowledge sharing and standard-setting

• Implemented a design token system across 100+ React components in Service Quoting & Service RO apps to support runtime CX/Interstate theme switching — orchestrated end-to-end with Claude Code agentic workflows
• Built a 41-prompt Claude Code library and CLAUDE.md constraint framework adopted as a spec-driven development standard
• Authored and maintains internal shared utility libraries: json-patch-utility · react-common-utility
• Led development of cloud applications — Service Quoting, Service RO, Pricing Engine, Catalog Manager — integrated across the full Xtime Spectrum suite (Invite, Schedule, Engage, Inspect)
• Delivered enterprise-grade SPAs for Citibank risk management, Wells Fargo real-time trading, and Walmart supply chain replenishment

• Centralized Agentic AI Knowledge Base — Cross-squad RAG pipeline ingesting SharePoint, Rally features & unstructured docs (pgvector · LangGraph · RAGAS · MCP)
• Architecture Decision Records (ADRs) — Formalizing key domain decisions for DMS+ Service as Principal-level proof artifacts
• Technical Roadmap — 12–18 month roadmap for Fixed Ops frontend architecture
• Certifications in progress — Anthropic Claude AI Courses · Google AI Essentials · GitHub Copilot

P5 promotion hinges on written evidence, not heroic delivery.

Deliver these proof points:

• 5+ Architecture Decision Records — formalize the big choices: why this pattern over that, what you traded away
• 12–18 month technical strategy — written roadmap for your domain evolution, tech debt, AI adoption milestones
• 2–3 cross-team RFCs — standardize agentic AI adoption, spec-driven workflows, or architecture patterns
• Knowledge base initiative — measurable adoption metrics (discovery time reduction, spike elimination)
• 8–12 technical articles — published on agentic AI, system design, modern frontend architecture
• AI generalist foundation — complete core coursework (Claude, LLM fundamentals, fast.ai); demonstrate breadth across the stack

Horizontal influence beats heroic delivery. The Principal who ships features everyone forgets is less valuable than the P4 who writes the RFC that unlocks three teams.

Written artifacts are proof. ADRs, RFCs, technical roadmaps, and published articles are how you show Principal-level thinking. Verbal influence is ephemeral.

Agentic AI and spec-driven development aren't trends. They're force multipliers — they scale through adoption and standards, not through headcount.

Generalist foundations before specialist depth. Understand the full AI landscape before you specialize in RAG or agents. The best specialists know what they're not specializing in.

Mentorship is a leadership multiplier. Your mentees shipping code they're proud of, advancing their careers, landing P5 conversations — that's the real proof of Principal-level impact.

IC path >> management path (for me). I ship code faster when I own the architecture. I mentor async via standards. I multiply team velocity through frameworks. That's where my leverage lies.

Frontend
React TypeScript Redux Node.js Sencha ExtJS AngularJS jQuery HTML5 CSS3

Backend
Java / J2EE Spring Hibernate REST SOAP JMS EJB Servlets

Agentic AI & ML
Claude Code LangGraph LangChain RAG pgvector RAGAS Prompt Engineering CLAUDE.md

Databases
Oracle / PL-SQL MS SQL Server MySQL IBM DB2

Cloud & DevOps
AWS Jenkins Maven Git Docker Tomcat JBoss Webpack

Testing & Quality
JUnit Selenium Puppeteer HP ALM TDD Code Health Scoring

Project & Collaboration
Rally JIRA Agile / SCRUM VS Code IntelliJ Eclipse

• Anthropic: Building with Claude series, Claude API documentation, agent architecture guides
• OpenAI: LLM engineering courses, prompt engineering best practices
• fast.ai: Practical deep learning for coders, LLM intro courses
• Maven Analytics: LLM engineering bootcamps, advanced prompt design

• Replit AI Bootcamp: Hands-on LLM application building
• Anthropic Advanced Courses: Multi-agent systems, evaluation frameworks
• LangChain / LangGraph: Official documentation and advanced tutorials

• Architecture Decision Records: ADR GitHub org, decision-making frameworks
• Test-Driven Development: TDD best practices, executable specifications
• System Design: URL shorteners, distributed caches, and real-world case studies

Lead Software Engineer · Cox Automotive · Frontend SME · Agentic AI · 18+ years total experience