Customer Development Interview. AI Cloud Compute Users

Customer Development Interview with AI cloud compute users

We are looking to speak with experienced AI practitioners who have hands-on experience using GPU cloud infrastructure for model training or inference.

This is a short research conversation about what has worked well and what has been painful in your past experience. The goal is to learn from practitioners and use those insights to shape a product in the future. It is not an evaluation of you, and is purely a learning conversation.

Who is a good fit?

You are:

• An AI Engineer, ML Engineer, Applied AI Researcher, or Technical Founder
• Currently working at:
  • An AI startup (Seed to Series B preferred), OR
  • An AI-heavy product company (gaming, video, agents, multimodal, LLM apps)
• Directly involved in infrastructure decisions for:
  • Model training (fine-tuning, SFT, LoRA, QLoRA, etc.)
  • Inference workloads (batch or real-time)
  • Long-running AI agents or multimodal pipelines

Infrastructure Experience Required

You have used at least one of the following beyond AWS/GCP/Azure:

• RunPod
• CoreWeave
• Lambda Labs
• Paperspace
• Vast.ai
• Modal
• Together.ai
• Any other GPU cloud provider

Bonus if youve:

• Switched providers due to pricing or reliability
• Experienced scaling issues across multiple GPUs
• Compared bare metal vs managed GPU solutions
• Faced GPU availability shortages

We are especially interested if:

• You manage AI compute budgets
• You care about price/performance optimization
• Youve struggled with unpredictable costs
• Youve deployed production inference workloads
• Youve optimized GPU utilization

Not a Fit If:

• You only used AWS Sagemaker once for a tutorial
• You have no direct infrastructure decision-making involvement
• You are not hands-on with model deployment

Research interview Details

• 30 minute structured interview
• Remote (Google Meet)
• Discussion topics:
  • GPU provider selection criteria
  • Pricing models and cost predictability
  • Performance bottlenecks
  • Workload types (training vs inference vs agents)
  • Switching costs and lock-in

To Apply

Please include:

1. What AI infrastructure providers have you personally used?
2. What type of workloads did you run?
3. Approximate monthly compute spend?
4. Your role in infrastructure decision-making?