Got an idea and a laptop? That's all you need. Team up, pick a track, and ship a real AI app on a full enterprise cloud — no setup, no gatekeeping, no experience required.
Enterprise infra · Live endpoints · Real users · Real outcomes.
Everyone's talking about AI apps. You could keep watching the timeline — or you could be the one shipping the thing it's about.
The Vayu AI Studio Hackathon is an end-to-end AI engineering competition built for college students, by people who believe your best project shouldn't be locked inside a university submission portal.
Pick a real problem. Pick a track. Build an AI application that does something useful for a real person — a farmer, a student, a clinic, a small business owner whose day gets meaningfully better because of what your team shipped.
Object storage, MLFlow, model registry, vector DB, MaaS catalogue, Kafka, Postgres, container registry, GitLab, realtime inference. Provisioned the moment you sign up.
Every track ships with a starter kit, a step-by-step build path, and a transparent deliverables checklist. You always know what to build next.
Tracks designed around real users across agriculture, healthcare, finance, logistics, and manufacturing.
Outcomes > aesthetics. A working, deployed app counts as much as the cleverest architecture.
MLFlow, RAG pipelines, agent graphs, human-in-the-loop. Skills that map 1:1 to industry.
Build a predictive ML app that helps a real user make a data-driven decision they couldn't make confidently before.
Build a computer vision app that automates visual inspection for a user who currently relies on manual observation.
Build a document intelligence copilot that answers questions grounded in a real corpus — with citations, in English and at least one Indian language.
Build an autonomous agent that takes a goal in plain language, orchestrates multiple tools, and completes a multi-step task on a user's behalf.
Build a physical AI system that ingests live sensor data, runs inference in the cloud, and delivers a command or alert back to a device or simulator.
Full briefs, sample use cases, deliverable checklists, and scoring rubrics are published on GitHub. Choose your battlefield before the bell.
View all 5 tracks on GitHub ↗Every track is scored on a common framework. Judging is fast, objective, and transparent — every criterion is either a binary checkpoint or a clearly defined band.
10 binary checkpoints verifying end-to-end platform usage — dataset uploaded, model registered, endpoints live, observability active.
Clarity of the user persona, the decision or task being addressed, and the before-and-after impact narrative.
Model accuracy, RAG faithfulness, agent robustness, or closed-loop reliability — evaluated against each track's specific requirements.
Functional stability, input validation, plain-language output, and the ability of a non-technical user to complete the intended task without help.
A 5-minute video demonstrating real usage, with a clear explanation of architecture and user value.
Multilingual support (Ask-It), human-in-the-loop safety gates (Do-It), or closed-loop operational proof (Move-It).
Complete scoring rubrics for each track are on GitHub.
View rubrics on GitHub ↗A Vayu AI Studio environment is automatically provisioned the moment your team leader signs up. Everything from here is execution.
Team leader signs up. A Vayu AI Studio environment provisions automatically — no setup, no DevOps.
Add 2–4 members via the Tenant dashboard. Shared workspace, repos, and credit pool.
Skim the five briefs. Pick the one that wakes you up. Switching later means re-cloning a kit.
Each track ships in your pre-created "Vayu Hackathon" GitLab repo. Fork or branch and begin.
Follow the guided journey. Hit the checkpoints. Record the video. Submit before the bell.
One per track. Best end-to-end deploy.
Projects that punched above their weight.
Keep building on Vayu long after the demo.