About Me

🤖 AI Solution Architect building enterprise agentic-AI platforms — 15+ years in architecture and engineering, 4+ years in GenAI, LLM, RAG, and multi-agent systems.

I’m Nagul Meera Mahankali — an AI Solution Architect and hands-on technical leader based in the Dallas–Fort Worth area, currently a Senior Vice President in AI Platform & Agent Engineering at JPMorgan Chase. I write AI Code Geek to share what I learn building agentic-AI systems at enterprise scale, in plain, practical language.

The Short Version

  • 🏗️ 15+ years in solution architecture, technology consulting, software engineering, and enterprise delivery
  • 🤖 4+ years designing and shipping GenAI, LLM, RAG, and agentic-AI solutions at global scale
  • 🏦 Currently SVP, AI Platform & Agent Engineering at JPMorgan Chase (Senior Lead Software Engineer)
  • ✍️ Founder of AI Code Geek and Liferay Savvy (250+ technical articles since 2012)
  • 📈 Building TradeAIQ — an experimental AI trading system, independent R&D on weekends
  • 🏀 Building SportCraft — another experimental weekend project
  • 🌍 Recognized Top Contributor to the Liferay open-source community, 2013–2016

My Journey

I started out deep in enterprise Java — Liferay portals, OSGi modules, Spring MVC, E-trading middleware — working across consulting engagements in India and the US for clients like Nintendo of America and Schneider Electric. That background in distributed systems, integration, and enterprise architecture turned out to be exactly the right foundation for what came next.

Since 2021 at JPMorgan Chase, I’ve been designing and building the firm’s enterprise agentic-AI platform — translating ambiguous business requirements into concrete solution blueprints, technical architectures, and MVP-to-scale delivery paths, then carrying them from concept through production with security, SRE, and governance partners.

What I Do Today

A few things I’m proud to have built:

  • A reusable MCP framework and developer SDK (built on Spring AI MCP) that became the internal standard at JPMorgan Chase — cutting time-to-market for production-grade agents by roughly 70%, and time-to-first-agent from weeks to days.
  • Multi-agent autonomous systems (LangGraph, LangChain, Claude Agent SDK, OpenHands SDK, Google AI SDK) that plan, build, test, and deploy code with minimal human intervention, adopted across multiple engineering teams.
  • Isolated, ephemeral execution environments for agentic CLI tools — AWS EKS containers behind a custom allow-list network proxy — enabling safe, governed, horizontally scalable agent adoption across a global developer group.
  • Responsible-AI and governance controls: a firm-level Skills/MCP registry, SSO-to-agent authorization brokering, a global kill switch, and secure MCP gateways that close the door on ungoverned “Shadow AI.”
  • A model-agnostic AI cost & budget platform giving leadership real visibility into spend and usage across LLM providers.
  • AI-powered remediation agents that autonomously identify, analyze, and fix source-code vulnerabilities at scale, plus a tiered SOC threat-triage agent fleet that auto-resolves roughly 80% of known threats — presented at the firm’s Global Hackathon.

Technical Toolkit

  • AI / GenAI / LLM: Agentic AI, Multi-Agent Orchestration, RAG & Vector Embeddings, Prompt Engineering, LangGraph, LangChain, Claude Code & Agent SDK, OpenHands SDK, Google AI SDK, Model Context Protocol (MCP), Spring AI MCP
  • LLM Providers: Anthropic Claude, Google Gemini, and OpenAI-based tooling — model-agnostic, multi-provider architecture with hybrid local + cloud routing
  • Cloud & DevOps / MLOps: AWS (EKS, S3, Secrets Manager, CloudFormation), Kubernetes, Helm, Terraform, Jenkins, GitHub Actions, CI/CD & GitOps
  • Architecture & Backend: Microservices, Event-Driven Systems (Kafka, RabbitMQ), Spring Boot, REST, API Gateway Design
  • Security & Governance: Zero-Trust, OIDC/JWT, RBAC, Threat Modeling, Governed Skills/MCP Registries
  • Languages: Java, Python, JavaScript/TypeScript, SQL

Beyond the Day Job

I’ve been writing publicly for over a decade. Liferay Savvy, which I started in 2012, grew to 250+ technical articles read by professionals worldwide and earned me recognition as a Top Contributor in the Liferay open-source community. AI Code Geek is the natural continuation of that habit — pointed now at GenAI, Claude Code, MCP, RAG, and agentic tool-calling — alongside an open-source portfolio of LLM tool-calling and vector-embedding architectures.

On weekends, I build TradeAIQ, an experimental AI trading system with predictors for swing, day, and intraday trading — my sandbox for testing ideas like an agent-state recovery framework that lets long-running autonomous agents resume where they left off after hitting token or context limits. SportCraft is another experimental project from the same weekend-builder habit.

My Philosophy

I believe in continuous learning, open collaboration, and hands-on experimentation. AI is transforming how software gets built, and I’d rather show my work — the architectures, the trade-offs, the things that didn’t work the first time — than just the polished outcome. My goal with this site is to make agentic AI accessible and practical, not just theoretical.

Let’s Connect

📌 Website: aicodegeek.com 📌 GitHub: github.com/ai-code-geek 📌 LinkedIn: linkedin.com/in/nagulmeeramahankali 📌 Email: meera.success@gmail.com

Join me on this journey as we unravel the wonders of agentic AI, one post at a time! 🚀