PROJECT TITLE:LapClusters

TAGLINE
Turn the laptops your team already owns into a private AI cluster for parallel code review.

CHALLENGE
Best Open-Source AI Project
Optional: Best Use of Gemma 4, if that challenge appears on our Coimbatore event’s challenge list.

PROJECT DESCRIPTION
Small teams often have to choose between reviewing code slowly on one local machine or sending it to a cloud AI service. LapClusters uses a team’s existing laptops as a private cluster for parallel code review.

The host collects supported source files from a local folder or Git URL and places one task per file into a Redis Streams queue. Each worker runs Gemma 4 locally through Ollama and pulls the next available task. The host combines structured findings into a Markdown report sorted by severity.

The local browser interface lets a team start a review, see connected laptops and the active file assigned to each worker, follow per-file progress, inspect findings, and download the report. Heartbeats and task recovery help work continue if a worker disconnects.

HOW WE USED GEMMA 4
Each worker sends its assigned file and review prompt to its own local Gemma 4 model through Ollama. The model returns JSON findings constrained to a schema. No hosted AI endpoint is used for inference.

WHAT MAKES IT DIFFERENT
LapClusters distributes independent review tasks across complete copies of a small model. It does not divide one large model across machines. Faster or newly available laptops can pull more work, and abandoned tasks can be recovered.

BUILT WITH
Python 3.11+, Redis 7 and Redis Streams, Ollama, Gemma 4 E4B, redis-py, httpx, python-dotenv, Docker, HTML, CSS, and JavaScript.

REPOSITORY
https://github.com/VishnuVardhanNk/LapCluster

DEMO VIDEO
[https://youtu.be/QDoHNfAD5zs?si=jhAWyCH7WKOksAnw]

HOSTED APP
Not applicable. The app runs locally on the host laptop; open the local interface at http://127.0.0.1:8765 while demoing.

TEAM
Team name: [ReLUactivation]
Members and contributions:

• [Nithin Krishnappa]: [Built the Redis Streams task queue and task status tracking]
• [Vishnu vardhan]: [Implemented the Ollama worker and Gemma 4 review flow, including structured findings]
• [Dheeraj]: [Built the local web interface for cluster status, review progress, and report downloads]
• [Niranjan]: [Integrated repository collection, host discovery, and worker recovery; prepared setup instructions and the demo]