Winner - MLH Hack Days Kanpur

Winning MLH Hack Days Kanpur with Agentic AI

From March 30 to April 2, 2026, I participated in the MLH Hack Days Kanpur, a competitive hackathon powered by Google Gemini and organized by AITD Events. Competing as a member of Team Syntrix, our objective was to build a viable, real-world application leveraging advanced Large Language Models within a strict 72-hour window. Our submission successfully secured the Winner title for the event.

The Event vs. The Project

1. The Event Experience

Hackathon: MLH Hack Days Kanpur
Organizer: Major League Hacking (MLH) & AITD Events
Duration: March 30 — April 2, 2026
Team Name: Team Syntrix
Outcome: Overall Winner 🏆

2. The Project Outcome: Architectural Foundation of "Sight Pro"

Instead of building a throwaway prototype, Team Syntrix focused on a real enterprise problem: massive bottlenecks in unstructured document processing.

My Role: I served as the Lead AI Architect, responsible for wiring the Google Gemini API to our Python backend and designing the asynchronous job queues to handle document parsing without dropping requests.
Constraints: We faced stringent API rate limits from standard LLM calls and high latency. To solve this within the hackathon timeframe, we engineered a custom caching layer and asynchronous job queues.
The Result: The prototype we pitched demonstrated an 80% reduction in processing time for unstructured enterprise data. This core architecture was later formalized and deployed as my standalone product, Sight Pro.

Timeline

March 30, 2026: Event Kickoff & Ideation. Team Syntrix finalized the focus on agentic data extraction using Google Gemini.
March 31, 2026: Backend Architecture. Built the initial Python/FastAPI server and integrated the Gemini RAG pipelines.
April 1, 2026: Frontend Integration & Optimization. Connected the Next.js dashboard and implemented the caching layer to bypass API latency constraints.
April 2, 2026: Final Pitch & Judging. Delivered a live demonstration of the document parsing workflow. Awarded Winner of the Hackathon.

Verified Credential Evidence

Award: Certificate of Achievement - Winner, Gemini-Powered Hackathon
Recipient: Team Syntrix
Issued By: Piyush Sahu (Organizer, AITD Events / MLH)

Key Learnings

Latency Management in Hackathons: Relying on live LLM APIs during a hackathon is risky due to rate limits. Building an async queue early on saved our final demo from crashing.
Agentic Viability: The event proved that local enterprises could immediately benefit from agentic workflows, inspiring the transition of our hackathon code into a production SaaS (Sight Pro).
Team Dynamics: Effectively splitting the workload—having dedicated focus on the Python cognitive layer versus the React dashboard—was crucial for delivering a complete stack in 72 hours.

FAQ

Did you build the entire Sight Pro application during the hackathon?

No. We built the foundational AI architecture and the core Gemini-powered RAG pipeline during the 72-hour MLH Hack Days. Following the win, I continued developing the platform into the enterprise-ready application it is today.

What technologies were used in the winning submission?

We utilized Python (FastAPI), Google Gemini API, React/Next.js, and implemented custom asynchronous job queues for rate-limit management.

Who was on Team Syntrix?

Team Syntrix is my competitive hackathon team. For this event, we divided responsibilities across AI architecture, backend development, and frontend UI/UX to ensure a comprehensive final pitch.