This week’s HP & NVIDIA Developer Challenge Office Hours gave participants a chance to ask questions, troubleshoot roadblocks, and share ideas as we head into the final stretch of the hackathon. Here's a quick recap of what came up:

️ Project Collaboration & Setup

One common hurdle: collaborating with teammates in AI Studio. By default, accounts are set up with one seat—but that’s not a hard limit. The team confirmed that if you email aistudio.support@hp.com, they can manually add a collaborator so your teammate can join your project.

Another hot topic was installation and compatibility. Participants asked about running AI Studio on services like RunPod and Linux-based systems. While not officially supported yet, the team encouraged folks to try installing the Linux version on Ubuntu 22.04 and share feedback. There’s even a helpful download and setup guide shared during the call.

Running AI Studio Without a GPU

A few users asked if AI Studio can run without an NVIDIA GPU. The short answer: yes, but with limitations. You can run some tasks—like image classification—on CPUs, but for LLMs or anything compute-heavy, it’s going to be slow. One user reported success using a CPU-only system with 64GB of RAM, but it’s not ideal.

If you’re not sure whether your setup meets requirements, the AI Studio quickstart page includes full hardware and software specs.

🧩 Third-Party Model Integration

Great news: You’re not limited to HP’s pre-built models. AI Studio supports third-party models via:

• Hugging Face libraries

• NVIDIA NGC

• Locally saved models through the interface

The team did note that not every model in the catalog is guaranteed to run without issues. If you run into bugs, use the built-in support tab in AI Studio to submit a ticket with logs attached. This helps the product team track and improve compatibility.

Use Case Discussions & Ideas

Participants shared potential projects, including skin disease progression tracking and marketing optimization. One community member offered a deep dive into applying Bayesian learning for A/B testing in marketing funnels and using computer vision in healthcare scenarios like melanoma detection.

Big takeaway: whatever your idea, clear documentation, model rationale, and well-defined metrics (like false positives/negatives) will help your submission stand out.

Quick Reminders

• There's still time to enter the sweepstakes to win an HP ZBook Fury G10—check Discord for Andrew’s post.

• The submission deadline for the hackathon is May 23, so make sure your project is on track!

Thanks again to everyone who joined this week’s session. The team is always looking for feedback—if you’ve got an idea or feature request, send it via the support tab in AI Studio or post it in Discord.

We’re getting close to the finish line—good luck and keep the questions coming!