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.
Detects bare-board PCB defects and converts detections into release/hold/scrap decisions and rupee rework exposure.
Predictive Maintenance for Railway Infrastructure. End-to-end IoT-to-ML pipeline for real-time railway track health monitoring, failure prediction, and automated crew dispatch.
Prove-or-abstain multilingual AI copilot for Indian government welfare schemes. Every answer is bound to the official line it came from -or the system refuses and escalates to a human.
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.
Register as a team, pick your track, and start building. Everything from here is execution.
Form a team of 2–4. The leader registers with every member's email, picks your AI track, and describes the problem you'll solve and the app you'll build.
15 teams are shortlisted on the strength of the problem you chose to solve and how you intend to build the capability for it.
Follow the guided journey. Hit the checkpoints. Record the video. Submit before the bell.
Upload your report, demo video, and live app URL. Deadline: 13th July 2026, 11:59 PM IST. Only shortlisted teams can submit.
Enter your registered team name and the email of the member submitting. Must match your registration exactly.
Project report (.pdf or .docx, max 10 MB) and demo video (.mp4, .mov, or .webm, max 100 MB).
Paste the public URL of your deployed solution. Add optional notes about your approach (up to 5,000 chars).
Build a predictive ML app that helps a real user make a data-driven decision they couldn't make confidently before.
Vayu Git ↗Build a computer vision app that automates visual inspection for a user who currently relies on manual observation.
Vayu Git ↗Build a document intelligence copilot that answers questions grounded in a real corpus — with citations, in English and at least one Indian language.
Vayu Git ↗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.
Vayu Git ↗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.
Vayu Git ↗Full briefs, sample use cases, deliverable checklists, and scoring rubrics are published on Vayu Git. Choose your battlefield before the bell.
View all 5 tracks on Vayu Git ↗Object storage, MLFlow, model registry, vector DB, MaaS catalogue, Kafka, Postgres, container registry, GitLab, realtime inference. Everything you need to build and deploy.
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.
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 Vayu Git.
View rubrics on Vayu Git ↗Second-best overall project.
Projects that punched above their weight.
Keep building on Vayu long after the demo — cloud credits to continue shipping on the platform.
Credits are valid till hackathon live date.