Amazon ML Challenge: Eligibility, Prizes, PPIs and Dates

Amazon ML Challenge is Amazon's national machine learning hackathon for engineering students in India. Teams get an Amazon dataset and a real problem, and have 72 hours to build a model. It is free, the top three teams win cash, and the top 50 teams get pre-placement interviews for Amazon's Applied Scientist Intern role.

Status on 30 October 2026: the 2026 edition is over. Registration closed on 22 September, the hackathon ran from 25 to 27 September, and the Grand Finale was held online on 7 October 2026. The next edition isn't announced. Registration opened on 26 August in 2024, 29 September in 2025 and 7 September in 2026, so expect August or September 2027.

Amazon ML Challenge at a glance

Eligibility and team rules

• Open to full-time PhD, M.E., M.Tech, M.S., MS by Research, B.E. and B.Tech students at engineering campuses in India.
• Two graduating batches per edition: 2024 took 2025 and 2026, 2025 took 2026 and 2027, and 2026 took 2027 and 2028.
• Teams of 2 to 4 in 2026. It was 3 to 4 in 2024 and 2025.
• Every team needs a team leader. Teams can mix colleges, but one student can only be in one team.
• In 2026 every participant had to create an AWS Builder Center profile first and enter its alias while registering. A wrong alias left the registration incomplete.

Stages and format (2026)

Along with the predictions file, each team submits a 1 to 2 page document on its approach, models and experiments, plus commented source code. Amazon ranks teams on both a public and a private leaderboard, so a model that overfits the public one can drop. You can only work from a laptop or desktop, and one login per participant.

Prizes and PPIs

• The top 50 teams get pre-placement interviews (PPIs) for the Applied Scientist Intern role at Amazon. This has been the same in 2024, 2025 and 2026. A PPI is an interview, not an offer.
• Top three teams also get certificates and goodies. The top 10 teams get certificates and Amazon swag, and 2026 added the same for the top 10 women-only teams.
• 2026 gave every registered participant $200 in cloud credits, and the top 500 teams at the 48-hour mark another $100.

Past editions

Amazon's write-up of the first edition doesn't print the year, but the batches (2022 and 2023) point to 2021. It drew 12,000+ registrations and 7,000+ submissions, and the task was classifying products into Amazon browse nodes. We couldn't find official dates for the 2023 edition.

What changed over the years

• Registrations on Unstop went from 74,823 (2024) to 82,787 (2025) and 89,392 (2026).
• The minimum team size dropped from 3 to 2 in 2026.
• 2026 added the AWS Builder Center step, cloud credits and awards for women-only teams.
• Early editions promised PPIs for "ML roles" to winning teams. Since 2024 it has been PPIs for the top 50 teams, for Applied Scientist Intern.

How to prepare

1. Form a team early, ideally with a mix of modelling and data skills.
2. The problems use Amazon product data (the first edition was product classification), so practise on large e-commerce datasets on Kaggle.
3. Get used to working on large datasets under time pressure and keep a log of every submission.
4. If you are new to ML, the free Amazon ML Summer School (registration usually May to July) covers the basics.

More contests that lead to jobs: hackathons and hiring challenges. Current Amazon openings and summer internships are on talentd.

Cover photo: Office building of Amazon at financial district of Hyderabad by Ishanjyotibora, CC BY 4.0, via Wikimedia Commons.