$ whoami_

Ganesh Bodke.
I architect systems
that think.

> _

Tech Lead, Solution Architect and AI Engineer with 8+ years designing distributed .NET platforms, cloud-native systems on Azure & AWS, and agentic AI. I embed with teams, decode ambiguous domain problems, and ship production systems fast - the full-stack, customer-facing engineer who owns problems end-to-end.

View work Get in touch Résumé ↗
scroll to explore
.NET Core / Microservices / Azure / AWS / Kubernetes / Terraform / Agentic AI / RAG / React / Next.js / CQRS / DDD / Event-Driven / Client Engineering / Rapid Prototyping / .NET Core / Microservices / Azure / AWS / Kubernetes / Terraform / Agentic AI / RAG / React / Next.js / CQRS / DDD / Event-Driven / Client Engineering / Rapid Prototyping /
about

Engineering lead, hands still on the keyboard.

GB Ganesh Bodke - Solution Architect & AI Engineer
Pune, India

I'm a Tech Lead and Solution Architect based in Pune, India, currently leading product engineering at Accelya. I own architecture decisions end-to-end while staying close to the code that ships.

My foundation is C# / .NET Core - microservices, Domain-Driven Design, CQRS, and event-driven systems running on Kubernetes across Azure and AWS. Lately I've gone deep on agentic AI: retrieval-augmented generation, vector databases, and LLM orchestration that grounds models in real domain context.

I thrive when I'm embedded directly with stakeholders and domain experts - decoding ambiguous problems, rapidly prototyping solutions, and shipping production systems that fit the real-world context. Whether it's an airline rule engine or a community platform, I adapt fast to unfamiliar domains and deliver end-to-end.

I care about systems that are resilient, observable, and simple to reason about - and about leveling up the engineers around me.

0
years engineering
0
cloud certifications
0
clouds: AWS + Azure
0
technologies shipped
experience

Where I've built things.

Sep 2023 - Present
Current

Tech Lead, Product Engineering

Accelya · Pune, India
  • Lead product engineering teams, owning architecture for a product catalogue platform and an AI-driven UI initiative.
  • Design microservices with clean architecture, DDD, CQRS, and event-driven patterns on Kubernetes.
  • Built an agentic AI proof-of-concept: RAG over a vector database with LLM orchestration grounded in domain data.
  • Work directly with airline domain experts and product stakeholders to translate ambiguous business problems into technical solutions - bridging the gap between what the business needs and what the platform can do.
  • Drive engineering quality - CI/CD on Azure DevOps & GitLab, SonarQube gates, JFrog artifact management, Terraform IaC.
Oct 2017 - Sep 2023
~6 yrs

Application Development → Senior Engineer

Accenture · India
  • Embedded with global client teams across industries - rapidly adapting to unfamiliar domains, codebases, and tech stacks to deliver production systems under tight timelines.
  • Built REST APIs and services on .NET Core with MS SQL and NoSQL data stores, containerized with Docker.
  • Practiced Agile delivery end-to-end - design, code review, automated testing, and production support - as the engineering partner clients relied on.
  • Foundation in cloud (Azure), DevOps tooling, and scalable backend design that shaped my architecture work today.
Education

Bachelor of Engineering (B.E.)

Savitribai Phule Pune University · Pune
  • Engineering degree underpinning a career in software architecture and distributed systems.
case studies

Architecture in practice.

Three systems from my work leading product engineering on a modern airline-retailing platform - how I approach AI, distributed systems, and platform design. Proprietary details are abstracted.

01 Agentic AI · LLM Orchestration · .NET Serverless

AI-Powered Rule Modernization

Role: Tech Lead & Architect - owned end-to-end across the AI backend, a new serverless data layer, and the UI.

The challenge

A business-critical airline rule engine lived on an aging XML datastore reachable only over SOAP - no search, and a desktop-era authoring tool that only specialists could use. The goal: let anyone author complex rules in plain English, safely.

What I built

  • An agentic AI service (FastAPI + LangGraph on AWS Bedrock) that turns natural language into validated domain XML.
  • A graph pipeline that generates JSON → renders XML → validates against the schema → self-heals invalid output in a repair loop - never a single blind prompt.
  • A .NET 8 serverless abstraction layer (clean architecture, CQRS) fronting the legacy SOAP engine, mirroring metadata to DynamoDB, documents to S3, and search to OpenSearch.
  • Human-in-the-loop approval so AI can never silently publish a rule.
Outcome

Plain-English rule authoring replacing a specialist desktop tool, on a searchable, governed, cloud-native foundation - with a resilient agentic pipeline that validates and repairs itself before anything is persisted.

Next.js UI natural language Agentic AI · LangGraph + Bedrock generate JSON render XML validate vs XSD self-heal loop human approval NxD Data Layer · .NET 8 serverless DynamoDB · S3 · OpenSearch Legacy engine SOAP / XML DB validated XML only
LangGraphAWS BedrockFastAPI.NET 8CQRSDynamoDBOpenSearchLambda
02 Event-Driven · Serverless · Clean Architecture

Event-Driven Product Catalogue Platform

Role: Architect & Lead - designed the serverless, event-driven service estate and its delivery model.

The challenge

Build the product-definition foundation for modern airline retailing - multiple services, owned by different teams, that must stay consistent, scale independently, and propagate changes in real time without tight coupling.

What I built

  • AWS Lambda + API Gateway microservices in clean architecture - one handler per operation, FluentValidation, DI throughout.
  • DynamoDB single-table design with deliberate PK/SK and full version history per entity.
  • EventBridge for asynchronous cross-service integration - config changes fan out to products as events, not direct calls.
  • Real-time partner push: DynamoDB Streams → EventBridge → orchestrator/worker fan-out; plus a two-phase, audited dev→prod promotion pipeline.
Outcome

A uniform, independently scalable estate with clean domain boundaries, real-time propagation, and auditable cross-environment promotion - the standard every new service now follows.

API Gateway→ Lambda DynamoDBsingle-table + versions EventBridgeDynamoDB Streams OpenSearchsearch / read model Partner Pushorchestrator → worker Env Promotiontwo-phase, audited Config servicesairline / system async events
.NET CoreAWS LambdaDynamoDBEventBridgeOpenSearchCQRSDDDTerraform
03 Platform Engineering · Micro-Frontends · Shared Frameworks

Shared Platform: Framework + Micro-Frontend Shell

Role: Platform Architect - drove the standards that make many teams' products look and behave as one.

The challenge

Many teams shipping many products meant drifting architecture and inconsistent UX. The fix wasn't another service - it was the platform underneath them all.

What I built

  • A shared .NET clean-architecture framework (Domain / Application / Infrastructure + an AWS layer wrapping DynamoDB, OpenSearch, EventBridge, Cognito, S3, Secrets Manager, SQS), shipped as a versioned package.
  • Every new service is built on it - so they're uniform by construction and onboarding is fast.
  • A micro-frontend shell (Next.js + Luigi) with Cognito SSO that composes each product into one experience.
  • A shared design system driving components and per-tenant branding via design tokens.
Outcome

Consistency, velocity, and governance across the estate - new services and UIs inherit the right architecture, security, and look-and-feel instead of reinventing them.

Browserone experience FLXOne Shell · Next.js + LuigiCognito SSO · product registry Catalogue UI FlxM AI UI API Hub UI DCS UI Falcon Design System - components + branding tokens Incubation .NET Framework - clean arch + AWS layerevery backend service builds on this all UIs share ↓
.NETNuGetNext.jsLuigi MFECognitoDesign SystemClean Architecture
personal projects

Beyond the day job.

Things I design, build, and ship on my own - under my studio Glannu Technologies. Real products, live in production.

Autonomous AI · Multi-Agent SystemLive · running daily

Job Agent - Autonomous Job-Application System

A multi-agent Python system I built from scratch that runs the entire job hunt end-to-end. It discovers roles across job portals and ATS platforms, scores each against my résumé with an LLM, tailors a fresh résumé per job using RAG, applies on my behalf via stealth browser automation, and reads, classifies, and answers recruiter emails - all from one command, every morning.

  • 5 specialized agents - discovery, match-scoring, RAG résumé tailoring, applier, and email - orchestrated through a job-status state machine.
  • Dual-LLM brain (Groq + Gemini) behind a unified wrapper with retry and rate-limit throttling; every response is structured JSON.
  • RAG over my résumé (sentence-transformers + vector store) so tailoring re-emphasizes real skills - never invents facts.
  • Stealth automation (Playwright + human-like behavior) with strict apply-confirmation checks and a dry-run safety gate.
  • Multi-user Streamlit app with auth, dashboards, and per-user workspaces; Gmail API for recruiter replies; SQLite tracking.
PythonMulti-AgentRAGGroq + GeminiPlaywrightStreamlitGmail APISQLAlchemy
Discovery agentportals + ATS scrape Match scoring · LLMJSON score + red flags Résumé tailor · RAGvector match → .docx Applier · Playwrightstealth + confirm check Email agent · Gmailclassify + auto-reply SQLite · status lifecycle
Web · Next.jsLive

Dike Patil - Public Figure Site

A bilingual (Marathi / English) website for a public leader: animated hero, a 40-year journey timeline, a category-filtered work showcase, a 57-image cloud gallery, press coverage, and maps + WhatsApp contact - all SEO-tuned.

Next.jsi18nSEO/OGImage Cloud
dikepatil.com ↗
Web · Next.jsLive

Rajeshahi Pratishthan - Org Site

A community organization site with year-based committee rosters, an events calendar, a multi-year journey timeline, and a mixed photo / video / reels gallery with Instagram integration. Deployed on Vercel.

Next.jsVercelMedia GalleryInstagram API
rajeshahipratishthan.com ↗
StudioLive

Glannu Technologies

My software studio - where I design and ship full-stack web products end to end: modern Next.js front-ends, cloud hosting, SEO, and AI-powered tooling for clients and my own ideas.

Full-StackNext.jsCloudAI
tech.glannu.com ↗
writing

Notes on building.

Field notes on AI engineering, software architecture, and shipping systems that hold up in production.

Jun 2026·7 min read·AI Engineering

Agentic AI doesn't need a bigger prompt - it needs a validation loop

How I turned an unreliable "the LLM generates XML" step into a self-healing pipeline that validates against a schema and repairs its own output before anything is ever persisted.

Read article →
Jun 2026·6 min read·Architecture

Event-driven by default: designing services that don't call each other

Why I reach for events over direct calls across a microservice estate - and the honest trade-offs an event-driven backbone actually buys you.

Read article →
Jun 2026·6 min read·Platform

One NuGet package made ten services look the same

The case for a shared clean-architecture framework - the velocity it buys a team, and the version discipline it quietly demands.

Read article →
Jun 2026•9 min read•Software Engineering

The Evolution of Code: Why and How Our Standard File Formats Were Invented

An architectural deep-dive into the origin stories, creators, and core data paradigms behind 15 of the most influential programming and configuration file formats.

Read article →
capabilities

The toolkit.

backend & architecture

  • C# / .NET Core expert
  • Microservices expert
  • CQRS · DDD · Event-Driven advanced
  • REST API design advanced
  • MS SQL · NoSQL advanced
  • Python proficient

cloud & devops

  • Microsoft Azure advanced
  • AWS advanced
  • Kubernetes · Docker advanced
  • Terraform advanced
  • Azure DevOps · GitLab advanced
  • SonarQube · JFrog proficient

ai & frontend

  • RAG · Vector DBs advanced
  • LLM Orchestration advanced
  • React · Next.js advanced
  • TypeScript · JavaScript advanced
  • Agentic AI systems building

client-facing & delivery

  • Stakeholder collaboration expert
  • Domain adaptation advanced
  • Rapid prototyping advanced
  • Cross-functional teams advanced
  • Agile · Jira expert
certifications
SA
AWS Solutions Architect - Associate
Dec 2025
DEV
AWS Developer - Associate
May 2024
AZ
Azure Developer Associate
May 2022
AZ
Azure Fundamentals
Aug 2020
contact

Let's build something.

Open to Forward Deployed Engineer, Solution Architect, Tech Lead, and AI Engineer roles - in India and abroad. The fastest way to reach me is email.

Email me LinkedIn ↗ GitHub ↗ X ↗ Chess ↗