The AI Multi-Domain Hackathon by INICAI is a comprehensive, multi-track competition designed to evaluate participants across diverse domains of Artificial Intelligence, including Computer Vision, Natural Language Processing (NLP), and Generative AI Systems. Unlike traditional single-task competitions, this hackathon simulates real-world AI engineering environments, where participants must design, optimize, and integrate intelligent systems under practical constraints such as scalability, latency, robustness, and interpretability.
Participants can compete in one or more of the following tracks:
- Computer Vision - ReID-X: A challenge to build a robust system for person re-identification (ReID) across multiple cameras. Given synchronized video frames from 7 overlapping camera views, the goal is to assign consistent global person IDs to every detected individual across all cameras and frames simultaneously. The core challenge focuses on association rather than detection.
- NLP - Multilingual Classroom Translation System: Participants develop an end-to-end pipeline that converts spoken English lectures into multilingual outputs, including translated subtitles and synthesized speech. The system must operate under strict real-time constraints with minimal latency while maintaining high accuracy and contextual integrity across various accents and background noise.
- GenAI 1 - DefenceRAG: This track challenges participants to build intelligent Retrieval-Augmented Generation (RAG) systems to understand and reason over complex defence policy documents. Models must interpret procurement manuals, financial delegation rules, and naval regulations to answer domain-specific questions with accurate reasoning and grounded references.
- GenAI 2 - AutoDeck AI: An individual, time-bound evaluation where participants build an end-to-end AI system that generates structured, high-quality presentations from prompts using RAG over a provided document corpus. The pipeline must retrieve relevant information, generate coherent slide content, and structure outputs into presentation-ready formats.
Each track emphasizes cross-domain generalization, system-level thinking, robustness to noisy inputs, and efficient deployment. Final hiring decisions are based on overall performance, innovation, and system design, considering approach, code quality, reproducibility, and practical applicability.
Top-performing participants will be evaluated for internship opportunities at INICAI and collaboration opportunities on INICAI-driven projects. Selection criteria include leaderboard ranking, but also heavily weigh qualitative factors such as model performance, system design, innovation, and code quality.