Artificial Intelligence
Tech Zephyr 4.0Agentic AI Hackathon
The Agentic AI Hackathon challenges participants to build intelligent systems capable of autonomous multi-step execution. Instead of simply generating answers, teams must demonstrate systems that can pursue goals, select actions dynamically, use tools, adapt to changing conditions and handle failures.
Prize Pool
₹35,000
Tap the stack to flip through
Registration
Ready to participate?
Confirm your registration slot for Agentic AI Hackathon. Complete your submission through the official registration form.
⚠️Only for participants shortlisted after Round 1
The Challenge
Highlights
Key Highlights
What Makes This Competition Different
Build Autonomous AI Agents
HighlightOnline Submission Round
HighlightLive Offline Challenge
Highlight7-Minute Final Pitch
HighlightExperience
What To Expect
Competition Experience
Built To Challenge You
Teams can select their own problem domain, including education, finance, healthcare, research, business operations, cybersecurity and scientific computing.
A successful agent should demonstrate goal-driven execution, dynamic action selection, multi-step execution, adaptation and robustness.
Teams may use LLM APIs, agent frameworks, RAG, vector databases, external APIs and public datasets.
Round 1 requires a problem statement, target users, proposed solution and explanation of why an agentic approach is necessary.
Teams submit architecture documentation, source code, demo video and a runnable implementation.
The grand finale introduces a brand-new challenge involving an Agentic AI chatbot.
Finalists present their solution for seven minutes followed by a five-minute technical Q&A.
Agentic AI Hackathon
Challenge · Build · Compete
Evaluation
How You Will Be Judged
Agentic Behaviour
CoreThe system must demonstrate meaningful autonomous goal pursuit and action.
Technical Depth
CoreQuality of architecture, tools, reasoning, memory and system implementation.
Robustness
CoreAbility to handle failures, unexpected conditions and changing inputs.
Impact
CorePractical usefulness and relevance of the chosen problem.
Schedule
Competition Timeline
Registration Deadline
Online Submission
Finalist Selection
Grand Finale
Rewards
What's At Stake
Guidelines
Rules & Regulations
Teams may contain one to four members.
Each participant may belong to only one team.
The proposed system must demonstrate genuine agentic behaviour.
Teams must submit a problem and solution brief.
An architecture diagram explaining the system is required.
Source code must be provided through a public GitHub repository.
A three-to-five-minute demo video is required for Round 1.
Finalists must be able to run the system live during the finale.
The final judging panel's decision is binding.
⚠️Only for participants shortlisted after Round 1