FloatChat

Overview

Built for Smart India Hackathon 2025 by a 6-member team that I directed — a team effort under a 36-hour national deadline.

Challenge

Deliver a working end-to-end AI system — ingestion, ML inference, API gateway and frontend — fast enough that six people could build in parallel without blocking each other.

Results

A functional live prototype demoed to national judges, delivered on deadline.

My Ownership

• Directed the 6-member team: scoping, task split, integration and demo
• Mapped end-to-end architecture across ingestion, ML inference, REST gateway and React frontend so all six developers could work in parallel across service boundaries
• Presented the functional live prototype to judges, walking through architecture decisions, execution tradeoffs and product flow

Technical Implementation

• Service boundaries drawn first — parallel work beat sequential perfection under a 36-hour clock
• Cut features early to protect the end-to-end flow for the live demo

Engineering Challenges

Leading means deciding what not to build. Integration time always costs more than planned.