Bharath Darapu

Bharath Darapu

Lead Architect & AI Engineer at Morgan Stanley

I design trading systems that move a million orders a minute, and build AI that rewrites legacy code into cloud services.

Scroll to watch six generations of agents. Each one is a stage of my career.

2012 to 2014

Assistant System Engineer

Tata Consultancy Services, then Syracuse University and an internship at ChronicleMe

  • WebBuilt modern web frameworks in the Enterprise Architecture Group.
  • MobileBuilt mobile and web features for an early-stage startup.

2015 to 2018

Associate, WM Operation Stability

Morgan Stanley

  • 100+Splunk and ExtraHop dashboards, and 30% less incident noise.
  • 150+engineers taught in monitoring and troubleshooting bootcamps.

2018 to 2022

Engineering Lead, Wealth Management Advisory

Vice President, Morgan Stanley

  • < 1 mindowntime per release with rail-based deployments, down from 6 hours.
  • 70%fewer production incidents after the stability program.

2022 to 2024

Lead Architect, Advisory Trading Solutions

Vice President, Morgan Stanley

  • 1M+orders per minute kept in sync between DB2 and SQL Server.
  • 1500%more efficient order routing with a new routing component.

2024 to now

Lead Architect & AI Engineer

Vice President, Morgan Stanley

  • JOLTTurns mainframe COBOL into cloud-native services. Highly Commended, Banking Tech Awards 2026.
  • KnowledgeCompassRAG platforms: engineering tasks 40% more efficient, decisions 25% faster.
  • MeastroA multi-agent framework for complex engineering tasks.
  • AI CivilizationMy side project, and the idea behind this page.

Message me on LinkedInSee my code on GitHub

Side projects

AI Civilization

LLM agents live full 10-decade lives inside a domain. They specialize, mentor, and argue as elders, and each generation passes its conclusions on. The output is a strategy report shaped by generations of dissent, not one model's best guess.

In design. Last changed 7 months ago.

Source on GitHub

ProjectBD for Android

A set of personal Android apps in Kotlin and Jetpack Compose. The fitness app tracks macros and workouts, scans barcodes with ML Kit, and syncs with Health Connect.

Kotlin. Last changed 1 month ago.

ProjectBD-FitnessProjectBD-DietProjectBD-Automation

ProjectBD Reddit

An ad-free Reddit client for Android, rebranded from an open-source fork.

Java. Last changed 1 month ago.

Source on GitHub

CORS Unblocker

A browser extension that turns off CORS checks in Chromium-based browsers during local development.

Last changed 7 months ago.

Source on GitHub

Also on GitHub

Skills

AI
Generative AI, RAG, multi-agent systems, agent benchmarking, AI code modernization
Architecture
distributed systems, microservices, REST APIs, high-throughput data sync, platform modernization
Cloud and operations
Kubernetes, OpenShift, Docker, CI/CD, SonarQube, Splunk, ExtraHop
Languages and data
Java, Kotlin, Python, C# and .NET, DB2, SQL Server
Mobile
Android, Jetpack Compose, Room, CameraX, ML Kit

This site

This site runs on a server in my home, next to my media, home automation, and disk monitoring services. A small Python program renders the page on every request, using only the standard library. The agents are drawn in your browser.

Also running there: Jellyfin, Plex, Home Assistant, the *arr apps, Nginx Proxy Manager and Cloudflare DDNS.

This page took 1.8 ms to render. The server has been up for 17 days, 3 hours.