TaskFlow: AI Project Management

A collaborative project management tool that uses machine learning to automatically predict task durations, assign resources based on team workload, and detect potential bottlenecks before they happen. Scaled to support over 10,000 concurrent active users with real-time websocket updates.

The Problem

Modern project management tools often require heavy manual input for estimating timelines and managing resource allocation. Teams end up spending more time managing the tool than actually doing the work. The goal was to build a system that passively learns from a team's historical velocity and auto-balances workloads.

Architecture & Tech Stack

The frontend is built with React and utilizes Socket.io for instantaneous collaborative updates (like Google Docs). The primary CRUD backend is a Node.js/Express monolith backed by PostgreSQL. The AI layer is handled by a separate Python microservice running FastAPI, which utilizes a trained RandomForest model to analyze historical completion times and predict future bottlenecks.

Impact & Outcome

TaskFlow was adopted by 5 internal teams initially, leading to a 30% reduction in time spent in sprint planning meetings. The websocket architecture successfully handled stress tests of up to 10,000 concurrent connections with under 50ms latency.