My reflections on the great vibes of the generative AI hackathon at University of California, Berkeley.
๐ Our project
scaledrive.ai - LLM for scalable roboticsย
We worked on integrating chatGPT / LLM with self-driving vehicles to help them better understand the context and plan a safer path accordingly. We only had a little time for preparation and started programming from scratch at the beginning of the hackathon.
๐ Results
1) Out of 1.800 applications, ~800 people submitted their projects, in total 268 teams. You can find all projects here https://lnkd.in/gPbQS3fe
2) Every team was supposed to be judged at least twice. The more judges came to a team, the closer it was to getting into finals. If I counted it correctly, seven judges visited us and heard our pitch, and most gave us positive feedback. We made it to the top 10%!
3) Some judges and mentors were investors and suggested connecting and discussing our project when we start fundraising.ย
4) I connected with many other incredible and smart people during the hackathon: engineers, investors, and AI enthusiasts who supported us and were interested in our project.ย
๐ My takeaways
1) Preparation is king. Before the hackathon, you should prepare as much as possible: discuss ideas and implementation, prepare frameworks and tools, assign tasks, read and understand rules, and judging criteria.
2) Make sure the team has met in person before the hackathon and that all members are aligned on goals, tasks, and ideas.
3) The number of work hours and lines of code during the hackathon matters. There were many brilliant people, and one team could distinguish itself from another by the quality of preparation and determination during the hackathon.
4) It is worth it. The final pitches were in front of renowned investors like Sequoia Capital and Lightspeed.
I am grateful to my team for their hard work and ideas. It was fun working with you during the hackathon, and considering limited preparation, I am proud that we managed to submit a meaningful project and get positive feedback from many judges.