Video Gallery

Recap02:07MantisGrid AI Hackathon & Summit 2026 HighlightsTwo minutes from the MantisGrid AI Hackathon & Summit 2026 in Palo Alto: teams hacking on AI infrastructure problems, the MantisGrid AI keynote and live product demo, panels with industry leaders, and the award ceremony.Watch now Keynote16:07Mithril on GPU Reliability at Scale | William VambenepeWilliam Vambenepe, VP of Product at Mithril, shares how Mithril is rethinking GPU infrastructure, capacity, and reliability in his keynote at the MantisGrid AI Summit 2026. Mithril sources GPU capacity from many providers, so it runs a complex, mixed infrastructure (1000s of GPUs) that needs consistent, high-quality operations. Mithril is a design partner and customer of MantisGrid AI, and we are working closely with them in three areas: → From alerts to automated RCA: full incident reports with root causes and recommended actions, so on-call engineers do less manual work → Recovering stranded GPU capacity: finding GPUs that aren't available on the market and bringing them back → Automated resolution: fixing incidents faster, while making sure real software or logic issues get a permanent fix at the source Thank you, William and the ‪@mithrilcompute‬ team, for the partnership and trust in MantisGrid AI.Watch now Keynote04:53How Rockfish Data Keeps AI Reliable with MantisGrid AI | Muckai GirishMuckai Girish, Co-Founder & CEO of Rockfish Data, gave a keynote at the MantisGrid AI Hackathon & Summit 2026 in Palo Alto on Sept 17, 2026. Rockfish Data helps enterprises make their data ready for AI. It is also a MantisGrid AI customer. Rockfish Data runs models and workflows for its own customers, and MantisGrid AI helps keep them reliable. In this talk, Muckai covers three areas: → Detecting problems early: spotting anomalies across live AI training and inference workloads → Finding the root cause fast: correlating signals to pinpoint what went wrong → Acting with human governance: moving step by step toward autonomous resolution, with people in control "When we test the product and then give (MantisGrid AI) feedback, suddenly the features get implemented rather quickly." – Muckai Girish Thank you, Muckai and the entire Rockfish Data team, for the design partnership and trust in MantisGrid AI.Watch now Keynote05:09How AccuKnox and MantisGrid AI Keep AI Secure and Reliable | Phillip PorrasPhillip Porras, Co-Founder & Chief Scientist of AccuKnox, gave a keynote at the MantisGrid AI Hackathon & Summit 2026 in Palo Alto on Sept 17, 2026. AccuKnox focuses on cloud and AI stack security. It is also a partner and a customer of MantisGrid AI. AccuKnox secures the workloads in MantisGrid AI's environment, and it is integrating MantisGrid AI for infrastructure reliability and operations. In this talk, Phillip covers three areas: → Telling a failure from an attack: a pager alert, rising latency, or unusual usage can mean either one → Two platforms, one picture: AccuKnox enforces security for AI services, and MantisGrid AI manages resources and finds the root cause → Resilience as one goal: security keeps systems running under attack, and reliability keeps them running when things break "Our goal here is to bring together security and resilience to provide overall trustworthy and resilient runtime operations." – Phillip Porras Thank you, Phillip and the entire AccuKnox team, for the partnership and trust in MantisGrid AI. Watch now Highlights02:54The Future of AI Infrastructure and AI Reliability | MantisGrid AI Summit Panel HighlightsHighlights from two panels at the MantisGrid AI Hackathon & Summit 2026 in Palo Alto on Sept 17, 2026. Panel: The Future of AI Infrastructure Moderator: Kandan Kathirvel, Founder & CEO, MantisGrid AI Panelists: Boris Scholl (Microsoft), Malhar Thakkar (Intel), Sam Aldrin (NVIDIA), Vijay Vusirikala (Arista Networks) Should we build more AI data centers? Boris Scholl of Microsoft says we still need more because the demand is there. But we also need to better use what we have, from GPUs to network and storage. Panel: AI Reliability Beyond the Model Moderator: Randolph Chung, Co-Founder & CTO, MantisGrid AI Panelists: Jared Quincy Davis (Mithril), Mahesh Ram (Solvvy), Vijay Krishnan (Turing.com) What are the biggest problems enterprises face in adopting AI? Mahesh Ram of Solvvy says the CXOs he meets are tired of hearing about capabilities and want provable outcomes. The systems to turn "a capability mindset to a reliability mindset" are not in place yet.Watch now Discussion05:53Build More AI Data Centers, or Use Them Better? | Boris SchollBoris Scholl (VP & Architect, Cloud & AI Engineering, Microsoft) answers three questions from Kandan Kathirvel (Founder & CEO, MantisGrid AI) on the "Future of AI Infrastructure" panel at the MantisGrid AI Hackathon & Summit 2026. More AI data centers? The demand is there, but we also need to get more out of the GPUs, network, and storage we already have. How MantisGrid AI helps shared customers: data sovereignty and business continuity are moving critical workloads back on-prem, onto mixed systems that aren't all one control plane or all Kubernetes. Five years from now: ubiquitous compute, ambient intelligence, and agents built around intent and outcomes instead of replacing if-then-else blocks.Watch now