Nebius Serverless AI Builders Challenge

Build with Serverless AI. Share with the community. Win.

The Nebius Serverless AI Builders Challenge is a community competition for ML engineers, applied researchers, and data scientists. Build a reproducible AI or ML project using Nebius Serverless AI Jobs, Serverless AI Endpoints, or both. Publish your code, explain your approach in a technical blog post, and help other practitioners learn from your work.

Whether you are running batch inference, building an evaluation pipeline, deploying an LLM endpoint, creating a RAG application, or automating an agentic workflow — we want to see practical examples that others can run, adapt, and learn from.

The Nebius Serverless AI Builders Challenge is in partnership with Nebius Academy.

Nebius Academy is the education and research hub of Nebius, connecting real engineering work with professional learning and applied AI research.

See the results

Why participate

Win up to $2,000 in Nebius credits

Plus badges, swag, and a featured spotlight on the Nebius blog and documentation.

Get hands-on experience with Nebius Serverless AI infrastructure

Publish a technical piece that showcases your work to the broader ML community

Earn a spot in the Nebius Awesome Serverless Projects library

What to build

AI & ML

Fine-tuning pipelines
Model serving endpoints
Evaluation harnesses
RAG pipelines
Data processing jobs
Agentic and multi-agent workflows
Benchmarking setups
MLOps
Classical ML (CPU only)

Healthcare & Life Sciences

Protein structure prediction inference
Genomic variant processing pipelines
Medical image segmentation endpoints
Clinical text classification jobs
Drug-target interaction scoring
Bioinformatics batch workflows

Physical AI & Robotics

Robot simulation batch jobs
Sensor fusion inference endpoints
Object detection and tracking pipelines
Trajectory prediction models
Sim-to-real transfer pipelines
Digital twin data processing

What to submit

Every valid submission requires the following:

Code repository

A public GitHub or GitLab repository containing:

Code that uses Nebius Serverless AI Jobs or Endpoints
A Dockerfile or reference to public Docker images
A README with setup instructions, hardware configuration, expected outputs, and approximate runtime or cost
An open-source license
No committed secrets or private data

Technical blog post

A post of at least 600 words, published on LinkedIn, Medium, dev.to, a personal blog, Hashnode, or equivalent, that:

Links to your repository
Is tagged #NebiusServerlessChallenge
Explains the problem, architecture, implementation, and results in your own words

Video walkthrough (optional, but recommended)

A short video (3 to 10 minutes) walking through your project: what you built, why you chose this approach, and how Nebius Serverless fits in. Think of it as a companion to your blog post — something that lets judges see your project in action and understand your thinking directly.

You can publish it on YouTube, Loom, or any publicly accessible platform, and include the link in your submission form.

How submissions are judged

Submissions are evaluated by a panel of domain experts and Nebius engineers across six criteria: technical implementation, reproducibility, educational content quality, product usage depth, real-world usefulness, and originality. Every valid submission is reviewed, and the highest-scoring projects in each domain and category receive awards.

For full details on how each criterion is evaluated, see the Terms and Conditions.

Judges panel

Sergei Nozdrenkov

Founder at Wildflow

Dmitriy Samsonov

Co-founder/Chief Scientific Officer at NormaCore

Sergey Arkhangelskiy

CEO and Founder at Positronic Robotics

Alex Batsuev

Co-founder and CTO at Normacore

Denis Sapegin

Principal Cheminformatics Engineer at Quantori

Challenge Results

Congratulations to our winners! Thanks to everyone who submitted!

Place

Winner or Team Name

Submission link

1st

Shmulik Avraham