Roam
Engineering Intern
San Francisco · On-site · Internship · Posted Sep 7, 2025
- Experience
- No specific overall years required; preferred: 1+ year full-time at a hypergrowth startup.
About the job
Build AI-powered game generation and 3D environment systems, optimizing real-time generation and improving games using player and self-play feedback.
What you'll do
- Architect LLM-powered game generation pipelines and agentic chains.
- Optimize game iteration, model accuracy, and generation speed across mobile and cloud; scale the product.
- Build feedback loops and context-aware creation features using multi-turn dialogue and fine-grained controls.
What they ask for
- Experience building and scaling products with agents, including multi-agent tool-calling, memory, planning, live evaluations, and observability.
- Experience with LLM post-training, preference data, data pipelines, and evaluation harnesses.
- Have launched products that grew to thousands of users and owned onboarding, growth loops, and retention.
- Experience with GPU optimization and CUDA kernels.
Nice to have
- Significant experience in software engineering and applied AI.
- At least 1 year of full-time experience at a hypergrowth startup.
Pay
$150 an hour (hourly compensation)
Tools
LLMs, CUDA
AI focus
Applied AI for 3D world and game generation, including agentic pipelines, LLM post-training, inference-time optimization, and feedback-driven testing and improvement.
Model-written summary · the original posting is authoritative