Six Edge AI-focused tracks — all running on fully local or hybrid. Connect to our HP ZGX Nano GB 10 and build real AI without cloud API limits.
Every project's AI inference must run on the event's ZGX Nano — not on personal laptops or cloud AI APIs. This is the core principle that makes SJSUHack different.
All teams connect to our on-site HP ZGX Nano powered with NVIDIA Grace Blackwell (GB10) Superchip ready for LLMs, image gen, AI agents, and local model serving. No personal laptop inference.
Data never leaves the room. Local inference means zero cloud exposure — better privacy, lower latency, no API costs.
Your AI inference must be local — but cloud tools for slides, video, and design (Canva, Figma, Loom) are welcome.
All projects are evaluated across four core dimensions. Track-specific rubrics may add additional criteria.
Novelty, creativity, and originality of the idea and execution.
Complexity, quality of code, and appropriate use of technology.
Real-world feasibility, potential impact, and scalability.
Clarity of pitch, quality of demo, and ability to answer questions.
Every project's AI model training and inference must run on the event's HP ZGX Nano GB10 — not on personal laptops or cloud AI APIs. This is the core principle that makes the Edge AI SJSUHack different.
Every team must submit all three deliverables by September 25, 2026 at 8:00 PM. All submissions via the registration form link provided at the event.
Public GitHub or GitLab repo with all source code, a clear README, setup instructions, description of how local/hybrid inference is implemented, and script to enable running on other machines.
A presentation covering your problem statement, solution, tech stack, local/hybrid architecture, and the impact of your project. Keep it concise and clear.
A sizzle video of your working prototype. Focus on showing how your solution solves the problem you set out to solve for the target audience it was built for. Share the problem, who it's for, the core feature solving the problem using the local AI model for inference, and your team name, tagline, and logo. Remember it's for the person that has the problem who may not be technical, so focus on showing how your project solves their problem. Max 2 minutes.
Submission deadline: September 25, 2026 — 8:00 PM. All three deliverables must be submitted via the registration form link provided at the event before hacking ends. Late submissions will not be accepted. The submission platform (Devpost or registration form) will be confirmed closer to the event.
Register now and decide which challenge you'll tackle at SJSUHack.