Core AI Developers

Building the intelligence itself

Build the intelligence behind tomorrow's applications.

Not Vibe Coding. This category is for people who train, fine-tune, evaluate, deploy and engineer AI systems — where calling somebody else's API is the starting point, not the answer.

Registrations close

Sunday, 20 September 2026

Competition begins

Thursday, 24 September 2026

Offline finale

Sunday, 18 October 2026

This is for you if

→You have trained or fine-tuned a model, not only prompted one
→You can explain why your evaluation harness is trustworthy
→You think about cost, latency and failure modes, not just accuracy

What it covers

Python and RMachine learning and deep learningLLMs, NLP and computer visionModel training and fine-tuningRAG and embeddingsAI agentsAI infrastructure and data engineeringEvaluation and deployment

Four rounds, from 24 September

Round 1 is individual. Everything after it is a team. Round 4 is offline.

Round 1 · Individual AI Technical Qualifier

IndividualOnline

A mix of coding and conceptual problems across Python, data processing, ML fundamentals, statistics, LLM concepts, prompt engineering, evaluation, architecture, RAG, embeddings, fine-tuning and debugging.

Judged on

Technical correctnessDepth of reasoningDebugging ability

Round 2 · Team AI Prototype

TeamOnline

Your team builds a functional prototype against an AI problem — a recommendation engine, a classification system, a RAG application, an agent, a predictive model, a vision application or an NLP solution.

You submit

Model and architectureDataset informationCodeEvaluation resultsDemoDocumentation

Round 3 · Advanced AI Implementation

TeamOnline

Improve the system you built for accuracy, reliability, speed, cost, scalability, hallucination reduction, safety and real deployment.

Judged on engineering quality. A thin wrapper over an external API will not score well here.

You submit

Evaluation deltasCost and latency analysisDeployment notes

Round 4 · Offline AI Grand Finale

TeamOffline

An advanced problem on the day: analyse it, prepare the data, design the architecture, build, test and present.

Judged on

Technical depthPerformanceInnovationDeployment practicalityCostReliabilityPresentation

Build the intelligence behind tomorrow's applications.

₹299 per participant. You register on your own — team formation opens after Round 1.