Technical Lead Manager - Training Runtime, Data(set) Movement
$380k - $500k • Remote • San Francisco • FullTime
Posted 4mo ago
Job Location
San Francisco
Tech Stack
Remote Work Policy
Fully remote
Employment Type
FullTime
Categories
Machine Learning Engineer
About the job
OpenAI is seeking a deeply hands-on Technical Lead Manager to own datasets throughout our training infrastructure. This role will define the future of how training jobs access data, encompassing APIs, storage contracts, versioning, benchmarks, debugging tools, and reliability guarantees to ensure consistent data access across current and future training frameworks. Initially, you will be the primary technical owner for dataset reads, working directly with code and collaborating with researchers, training framework owners, storage teams, and infrastructure partners to build a durable platform. The challenge lies in making enormous, heterogeneous datasets easy to consume, correct across distributed workers, observable, and flexible enough to support various training modalities at frontier scale. The long-term vision is to build a team responsible for fast, correct, scalable, and reliable in-cluster data movement for training, expanding ownership beyond datasets to include checkpoint loads/saves and snapshot transfers.
Responsibilities
- Design and build a unified dataset read platform for multiple training frameworks.
- Define dataset APIs, storage-format expectations, registration/versioning, and migration paths.
- Build reliability into the read path, including stateful iteration, caching, fast restart, recovery, and operational contracts.
- Develop visualizers for inspecting text, multimodal, and reinforcement learning data.
- Write and review production code for data loading, service, caching, and reliability.
- Partner with teams working on training frameworks, reinforcement learning, multimodal models, storage, runtime, and cluster infrastructure.
Requirements
- Experience building or owning dataset, data loading, storage, or distributed training infrastructure at large scale.
- Proficiency in API design, debugging ergonomics, performance, and bit-level correctness.
- Understanding of failure modes in large distributed training jobs and how data systems impact them.
- Experience with stateful iterators, checkpoint/restart semantics, caching, remote services, or high-throughput storage reads.
- Comfort working across Python and lower-level systems code (Rust or C++ experience is a plus).
- Experience with multimodal, video, reinforcement learning, or pretraining data pipelines.
- Ability to lead through code and technical judgment, and manage engineers while remaining hands-on.
- Focus on developer experience by eliminating friction in data access and preprocessing.