Boda Madhukar Reddy: I build fast .NET systems, and ship them with teams of AI agents.

Senior backend engineer at Revalsys, Hyderabad. I lead AI coding agents (Claude Code, Codex, Copilot, Antigravity) with specs, skills and review, and take what they build to production: multi-tenant RAG and MCP platforms, Solr 9 hybrid search, and APIs load-tested with k6.

01RAG latency reduction
40s → 13s
02MongoDB query tuning
5–50 ms
03Agent-built foundation
2 days
04Building since Dec 2021
4 years and 10 months of experience, since 27 December 2021

( The journey so far )

Nearly five years, one thread: make it work, make it fast, then teach machines to help.

  1. Chapter 00: Prologue · Karimnagar

    It started with a B.Sc

    Computer science at Satavahana University, Karimnagar: data structures, algorithms and the first programs that actually ran. Enough to know I wanted to build things people use.

  2. Chapter 01: Dec 2021 · Hyderabad

    First job, real traffic

    I joined Revalsys and went straight into production code: core features for high-traffic e-commerce brands such as Jockey, Speedo, Manyavar and Fossil India on Angular and .NET, the integration with ONDC, the Open Network for Digital Commerce, and the money and parcels behind every order: Juspay, Paytm and PayU payments, GoKwik checkout and Shiprocket shipping.

    5 e-commerce brands · ONDC · payments and shipping

  3. Chapter 02: Making it fast

    Learning to measure before guessing

    Slow pages taught me to measure first. I built repeatable k6 load tests with Grafana dashboards to set real SLA baselines, and traced slow queries to full collection scans, fixed with the right MongoDB indexes.

    500–2,000 ms → 5–50 ms queries

  4. Chapter 03: Leading R&D

    From prototypes to systems

    I became the sole lead for the company’s .NET R&D: moving legacy systems to modern .NET, a WebRTC video platform, IoT proofs of concept over MQTT, offline-first apps and an incident-management platform. Most of them started as a question and ended as something running.

    Every .NET R&D project, one lead

  5. Chapter 04: AI arrives

    Building what AI runs on

    Then the work turned to AI: a multi-tenant RAG platform on .NET 10, an MCP server that gives assistants a map of C# code and SQL Server schemas, hybrid search on Solr 9, and models moved onto our own GPU server instead of a cloud bill.

    40+ MCP tools · ~₹10,000 a month saved

  6. Chapter 05: Today

    Leading a team of AI agents

    Now I lead AI coding agents the way a tech lead leads people: Claude Code plans and delegates, Antigravity builds, Copilot and Codex review, and every change is proven before it ships. The same discipline took our RAG chatbot from 40 seconds an answer to about 13.

    40 s → 13 s answers · 260 person-days → 2-day foundation

( From the desk )

The writing

All 17 posts
Boda Madhukar ReddySenior Software Engineer · AI Architect · Agentic Full Stack Developer

Boda Madhukar Reddy

Senior Software Engineer · AI Architect · Agentic Full Stack Developer Hyderabad, India

I build fast .NET systems, and ship them with teams of AI agents.

Senior backend engineer at Revalsys, Hyderabad. I lead teams of AI coding agents (Claude Code, Codex, Copilot, Antigravity) with specs, skills and review, and take what they build to production: multi-tenant RAG and MCP platforms, Solr 9 hybrid search, and APIs load-tested with k6 before they ship.

RAG latency reduction
40s → 13s
MongoDB query tuning
5–50 ms
Agent-built foundation
2 days
Building since Dec 2021
4 years and 10 months of experience, since 27 December 2021

From the blog

Latest writing

LatestArtificial IntelligenceNovember 18, 20263 min read

When Vector Search Wasn't Enough: A Table of Contents per Document, Searched in Parallel

Our RAG assistant indexed every document and still gave answers that missed the point. What fixed it: an LLM writes a table of contents for each document at indexing time, and at question time an LLM picks sections from those contents while vector search runs in parallel. Both result sets are merged and re-ranked before the answer is written.

RAGLLMspgvectorDocument AI
Read
Artificial IntelligenceNovember 14, 20267 min read

My AI Dev Team Lives in Herdr: Claude Code as Tech Lead, Antigravity as Developers, Copilot and Codex as Reviewers

Nine agent sessions, one owner. Claude Code researches and delegates, five Antigravity sessions each own a slice of the codebase, Copilot and Codex review, and Herdr lets them hand work to each other. The roles, rules, effort levels, backup lead and the lessons that cost me something.

Result: 9 agentsone lead, two reviewers

Claude CodeAI AgentsHerdr
Read
Artificial IntelligenceNovember 13, 20266 min read

Why I Call Myself an Agentic Developer: The Evidence, With Links

Everyone is agentic this year. Here is the case file instead of the claim: agentic products, the MCP tools agents use, the agent teams I run, and the numbers behind each, from 40 s to 13 s answers to a 260 person-day plan with a working foundation in two days.

Result: 105 / 166commits made with an agent

AI AgentsClaude CodeMCP
Read
Artificial IntelligenceNovember 12, 20266 min read

One Lead, Three Agents: Building an On-Prem AI Email Assistant with Claude Code, Codex, Cursor and Antigravity

A 260 person-day plan, and in two days a working foundation in 110 commits. How a spec, one rule book, a team file and a lead agent kept four coding agents in line, and how we measured the local AI instead of trusting it.

Result: 260 PD → 2 daysto a working foundation

Claude CodeAI AgentsOllama
Read
Artificial IntelligenceNovember 11, 20268 min read

Two Days, Not Two Sprints: Migrating Worker Services to .NET 10 with Skill Files and Coding Agents

Legacy background services moved to cross-platform .NET 10 worker services in two days. The coding agent did the typing; two stacked skill files, a house style and a migration playbook, told it what done looks like.

Result: 2 days.NET 10, every platform

.NET CoreBackground ServicesSkills
Read

Browse

Topics

All topics