Skip to content

Team · Hackathon Team Project | GenAI Product Prototype

Multimodal AI Virtual Assistant

1st Place · Hack-AI-thon (KARE ACM & Diebold Nixdorf) · 39+ teams · ₹5,000 · Dec 2024

An award-winning multimodal generative AI assistant built during a 24-hour hackathon, unifying voice, visual, and text workflows into a functional real-time assistant that won 1st place among 39+ competing teams.

My contribution

As a core team member at KARE ACM & Diebold Nixdorf's Hack-AI-thon (Dec 2024), I contributed to GenAI solution shaping, rapid prototyping, integration, and the final live demonstration. Judging recognised the team delivery, not an individual award.

Generative AI · Multimodal AI · LLM orchestration · Rapid prototyping · Python · AI product development

Problem

Standard text-only chatbots create friction in fast-paced, multi-sensory workflows where users need seamless context switching across input and output modalities.

Solution & architecture

Designed and demonstrated an end-to-end multimodal GenAI pipeline within a strict 24-hour sprint. Integrated LLM orchestration with rapid inference endpoints to deliver responsive, context-aware interactions under hackathon constraints.

System pipeline

Input ingestion and preprocessing → multimodal routing and context assembly → LLM inference → response synthesis. Engineering trade-offs prioritised low-latency inference and deterministic fallback handling so the demo stayed stable under live judging.

My contribution

Core-team contribution across solution design, prototyping, integration, and demo. Attribution remains collaborative.

Impact & recognition

Secured 1st place (₹5,000 cash prize) out of 39+ teams for turning a time-boxed GenAI concept into a stable, live-demonstrable product.

Visuals

Concept visual of a multimodal assistant with text, image, audio, and video panels.

Concept visual

Concept visual. Original illustration of the hackathon concept.