Member of Technical Staff - Data & ML Infra Engineer

Introducing Moonlake, AI for creating world simulations.

Moonlake is building the frontier of interactive world models: systems that generate, simulate, and reason over 3D environments for robotics, embodied AI, simulation, gaming, and multimodal agents. We are developing AI-native tooling for creating dynamic worlds where geometry, physics, visuals, audio, and interaction remain coherent in real time.

Our team sits at the intersection of:

• Embodied AI

• Robotics simulation

• Interactive 3D worlds

• World models

• Real-time generation

• AI infrastructure

Moonlake is building the next generation of AI infrastructure for interactive digital worlds. Our mission is to enable anyone to create, simulate, and interact with rich environments using natural language and multimodal inputs, turning simple ideas into worlds with structure, logic, and agents that can perceive and act.

Our team has raised $28M in seed funding from NVIDIA Ventures, Threshold Ventures, AIX ventures and notable angels including Naval Ravikant and Jeff Dean to build the foundational layer for the future of AI - powering everything from creative tools and games to robotics training, simulations, and digital twins. Our goal is to make building and experimenting with these environments as accessible and scalable as publishing video on the internet.

We are looking for exceptional research engineers and applied researchers to help push the frontier of interactive AI.

We’re looking for a Member of Technical Staff — Data & ML Infrastructure Engineer to help build and optimize the systems that power Moonlake’s model training and inference infrastructure.

This role sits at the core of Moonlake’s platform and focuses on one mission:

Improve throughput, latency, and cost — deploying models 2–10× faster and cheaper without quality regressions.

You’ll work across GPU kernels, inference systems, distributed training, serving infrastructure, observability, and large-scale orchestration systems.

This is a highly technical systems role intended for engineers who enjoy operating at the intersection of:

• ML systems

• Distributed infrastructure

• GPU optimization

• Production AI deployment

• Performance engineering

This role emerged directly from Moonlake’s need to better support large-scale world-model training and deployment infrastructure.

• →

  Optimize large-scale model training and inference systems

• →

  Improve GPU utilization, latency, throughput, and deployment efficiency

• →

  Build infrastructure that supports real-time world-model and multimodal workloads

• →

  Develop and optimize serving pipelines for frontier AI systems

• →

  Work closely with research teams to productionize high-performance models

• →

  Build scalable orchestration and observability systems for distributed AI infrastructure

• →

  Improve reliability, rollout safety, autoscaling, and production monitoring

• →

  Design systems that support fast experimentation without sacrificing stability

GPU Performance Optimization

• CUDA / Triton kernels

• FlashAttention family

• Paged attention

• CUDA Graphs

• Memory optimization

• Kernel-level performance tuning

Model Serving & Inference

• TensorRT-LLM

• Triton Inference Server

• vLLM / TGI

• Continuous batching

• On-GPU KV cache reuse

• Speculative decoding / Medusa

• Mixture-of-agents routing

Distributed Training & Parallelism

• FSDP / ZeRO

• Tensor parallelism

• Pipeline parallelism

• Expert parallelism

• NCCL tuning

• Multi-node GPU orchestration

Quantization & Efficient Fine-Tuning

• AWQ / GPTQ / FP8

• LoRA / DoRA serving

• Efficient deployment pipelines

Infrastructure & Systems

• Ray

• Kubernetes

• Argo

• Autoscaling systems

• Canary deployments & rollback infrastructure

• A/B experimentation systems

• Observability stack:

  • Prometheus

  • Grafana

  • OpenTelemetry

Moonlake’s products require real-time, highly efficient AI infrastructure capable of powering interactive worlds and embodied intelligence systems at scale.

The difference between:

• 200ms and 2s latency

• 40% and 90% GPU utilization

• Stable rollout and catastrophic regression

…directly impacts the company’s ability to train, deploy, and scale world-model systems.

You’ll help define the infrastructure foundation behind the next generation of interactive AI systems.

We are committed to being an on-site, in-person team currently based in San Mateo