# Prasanna Vadivelu — AI & Platform Product Manager

> Prasanna Vadivelu is an AI and platform product manager based in Herndon, Virginia, in the Washington, DC area. He has worked in product for more than 12 years, across AI platforms, e-commerce and enterprise software, after starting as a full-stack developer. His work includes enterprise learning and Microsoft Teams integrations at Microsoft, e-commerce serving about 25 million monthly sessions at Samsung, and Elly, an AI communications platform, at C1. He builds to test product ideas: Smart Route, his current project, coordinates several AI coding agents with defined roles, shared context and human review.

Based in Herndon, Virginia (Washington, DC area). Works across time zones with product and engineering teams in the United States, Canada, Europe and India.

## Roles
- AI Product Manager
- Platform Product Manager
- Technical Product Manager

## Experience
- C1: Elly, an AI communications platform (contact-center AI)
- Samsung Electronics: E-commerce at about 25 million monthly sessions
- Microsoft: Enterprise learning (LMS) and Microsoft Teams integrations
- Started as a full-stack developer before moving into product leadership.

## Focus areas
- AI product management
- Platform product management
- Technical product management
- AI agents and multi-agent orchestration
- Human-in-the-loop AI systems
- LLM evaluation design
- Local and on-device AI
- Developer platforms and tools
- Contact-center AI (CCaaS)
- Enterprise SaaS integrations
- E-commerce at scale
- Product strategy and roadmaps
- Full-stack software development
- Real-time 3D and interactive prototyping

## Current work: Smart Route
Smart Route is a local-first workspace that has several AI coding agents (Claude, Codex, Cursor, Kimi and others) work as one team. An intake step turns a rough request into a clear brief and proposes a team; a coordinator assigns roles (plan, build, review, recovery); decisions and context travel with each task; and human review remains explicit at the approval and integration points. It is built as a native SwiftUI workspace, a TypeScript coordination service and provider adapters.

The product questions it explores: when is multi-agent coordination worth its overhead, how should work be handed between models, and where must a person stay in the loop. The evaluation design compares local-only, frontier-only and mixed teams on the same bounded tasks by verified completions, elapsed time, model cost, rework and recovery. (An evaluation design, not a published result.)

About 70% of the prompts behind the interactive playground came from his daughter as short, plain requests. Smart Route had to ask the follow-up questions and turn them into buildable goals.

## How he works
Frame the problem. Build to learn. Verify what changed. He prototypes to test product decisions, writes down trade-offs, and separates evaluation plans from claimed results.

## Pages
- [Home](https://prasanna.id/): Smart Route, how it works, and the playground
- [About](https://prasanna.id/about): an interactive 3D recreation of his desk
- [Smart Route engineering story](https://prasanna.id/experiments/smart-route): architecture, decisions and trade-offs
- [The Smart Route team](https://prasanna.id/team): the roles in the agent workflow
- [Playground](https://prasanna.id/play): interactive simulations built with Smart Route (Moon, Mars, launch, solar system)

## Contact
- LinkedIn: https://www.linkedin.com/in/prasanna-vadivelu/