Modal

Serverless GPU compute built for AI/ML workloads and long-running agent loops.

Overview

Modal lets you deploy Python functions and web apps that automatically spin up GPUs or high-memory CPUs on demand. It's optimized for ML training, inference, and agentic workflows that require persistent state or long execution times.

Best for

• Python-heavy AI agents needing GPUs
• Long-running model fine-tuning or batch inference

Trade-offs

What XeroHack pre-wires

When you pick Modal in the interview, the engine emits these files into your scaffold:

• FastAPI + vLLM
• LangChain async agents
• PyTorch training jobs