Member of Technical Staff, Model Training Infra Platform
The work
- Architect and scale infrastructure for supervised fine-tuning, reinforcement learning, and on-policy distillation across language, multimodal, and robotics workloads.
- Design pipelines that coordinate GPU workers, rollout generation, teacher-model inference, reward computation, and simulation environments.
- Improve GPU utilization, memory management, communication, storage, scheduling, and distributed execution performance.
- Build agent runtimes, tool interfaces, environment integrations, model registries, and serving APIs.
- Support experiment tracking, dataset and model versioning, checkpointing, fault recovery, observability, and reproducibility.
- Identify infrastructure bottlenecks and reduce experiment turnaround time.
- Define platform interfaces and architectural patterns, lead initiatives from design through operations, and mentor engineers.
Requirements
- Language: Professional fluency in English.
- Required foundations: distributed systems, concurrency, networking, performance engineering, and fault tolerance.
- Required experience: designing, building, and operating scalable backend systems or distributed computing platforms; GPU-based machine learning infrastructure; distributed training; model inference; large-scale compute orchestration; and reliable APIs, services, and infrastructure for complex engineering or research workflows.
- Required technical skills: Python and at least 1 systems-oriented language such as C++, Rust, or Go.
- Technical ownership and sound architectural judgment in evolving research environments are important.
- Skills: Distributed ML Training Infrastructure, GPU Performance & Inference Optimization, Distributed Systems & Platform Architecture, Technical Leadership & System Ownership
Open to applicants in
- Argentina
- Austria
- Canada
- Chile
- Colombia
- Egypt
- France
- Germany
- Ghana
- Greece
- India
- Italy
- Japan
- Jordan
- Kyrgyzstan
- Lebanon
- Malaysia
- Mexico
- Morocco
- Netherlands
- Oman
- Pakistan
- Philippines
- Portugal
- Qatar
- South Africa
- Spain
- Sri Lanka
- Tunisia
- United Arab Emirates
- United Kingdom
- United States (Your country)
- Vietnam
Preferred
- Helpful background includes PyTorch Distributed, DeepSpeed, Ray, Megatron-LM, vLLM, TensorRT-LLM, CUDA, NCCL, Kubernetes, cluster scheduling, cloud GPU infrastructure, simulation platforms, robotics infrastructure, agent evaluation systems, and open-source distributed computing projects.
Build robotics hardware integrations, sensor pipelines, teleoperation systems, and video streaming infrastructure in an English-language contract role paying $67.31 to $115.38 per hour.
Build production LLM applications, RAG systems, and multi-agent tools in a full-time hybrid contractor role near Washington, D.C. Pay is $70 to $200 per hour, with applications starting through OpenTrain.
Build full-stack web applications for AI training workflows using React.js, Node.js, AWS, PostgreSQL, Serverless Framework, and WebRTC. This contract role pays $60 to $110 per hour.
Build and improve Go software used in AI training, including application development, code reviews, performance work, and technical documentation. This remote contractor role pays $60 to $110 per hour and requires a computer science degree.
Build AI-powered security systems that find, prioritize, and help fix vulnerabilities across applications, infrastructure, and AI workloads. This contract role is open to English-fluent candidates in listed countries.
AI & Security App Security Tooling LLM Security Programming Proficiency