Senior ML Researcher — Large Models, On-Site SF

Raydar; San Francisco (CA); On-site; USD 200,000 - 300,000; Full time

Benefits offered by this job

Equity

Health, medical and dental coverage

Job summary

Raydar is recruiting for a senior research role in San Francisco, leading research from hypothesis to working model and tying offline experiments to real-world results. You will work across model training, data pipelines and evaluation design, partnering with domain specialists to turn findings into robust systems.

The ideal candidate has hands-on experience training large models and a track record of turning unclear questions into production-ready solutions, while mentoring colleagues and

Qualifications

• Proven track record building and refining large neural networks with clear results.
• Mastery of a core ML technique (RL, LM scaling, data generation, or world models).
• Fluent Python-based deep learning coding skills.
• Hands-on experience with distributed training frameworks.
• Ability to design repeatable tests and guard against shortcuts.
• History of bringing open-ended research to production-grade models.
• PhD studies count toward experience; 5–10 years in research/engineering preferred.
• Ability to quickly learn unfamiliar technical domains.

Responsibilities

• Modify and train large models from base-training to instruction tuning.
• Build RL methods that learn from real-world outcomes.
• Design synthetic-data pipelines with provenance and quality controls.
• Create benchmarks to generalize to new problems and relate offline to real-world performance.
• Run and troubleshoot large distributed training jobs and optimize costs.
• Analyze learned models for prediction, search, and planning with strong baselines.
• Collaborate with domain specialists and engineers to prioritize improvements.
• Help set research agenda, raise engineering practices, and share work via papers/open-source.

Skills

Large NN development

Core ML technique mastery

Python programming

Distributed training

Experiment tooling

Research-to-production

Experience 5-10 years

Rapid learning

Education

PhD or doctoral study counted toward experience

Tools

FSDP

DeepSpeed

Megatron

Job description

Raydar is recruiting for a senior research role in San Francisco, leading research from hypothesis to working model and tying offline experiments to real-world results. You will work across model training, data pipelines and evaluation design, partnering with domain specialists to turn findings into robust systems.

The ideal candidate has hands-on experience training large models and a track record of turning unclear questions into production-ready solutions, while mentoring colleagues and