SkyDeck Dinner Series: Minima x Berkeley Grad Students

​Berkeley grad students: come have dinner with the founders of Minima AI (Berkeley SkyDeck Batch 22), the team making large language models run 2-10x faster on about half the hardware, fully on-prem or in a private cloud. Their stack combines sensitivity analysis, tensor-network compression, custom Triton kernels, and speculative decoding to accelerate open-weight or custom models with only a 1-5% accuracy delta and no application changes.

​

​Best part: it's a small dinner with founders building a real product, so bring your own models, workloads, or open-source projects and talk through running them faster on Minima. If you work on ML systems, model compression, GPU kernels, or efficient inference, you'll want a seat at this table. Co-founders David and Sergii (ex-Principal AI Engineer at Atlassian) love the technical weeds, and dinner is on us. Spots are limited, so RSVP below. Learn more at mnma.ai.