Safety Evaluations Platform Engineer – Secure Infra & Data

B Capital; San Francisco (CA); On-site; USD 180,000 - 240,000; Full time

Benefits offered by this job

Top-tier compensation

Stock options

Health & wellness

Meals provided

Parental leave

Unlimited vacation days

Visa sponsorship

Job summary

Reflection is seeking a Research Software Engineer on the Safety team to design, build, and own infrastructure for evaluating open models in high-consequence domains. You will create sandboxed environments, secure data pipelines, and robust tooling to support trusted experimentation.

You will collaborate with safety researchers, legal, and domain experts, translating evaluation needs into scalable, auditable systems with strong access controls and encryption at rest and in transit.

Qualifications

• Strong software engineering skills, particularly in Python.
• Experience building sandboxed, isolated, or security-sensitive execution environments.
• Solid grounding in security engineering fundamentals: least privilege, need-to-know, RBAC, secrets management, encryption, audit logging, defense-in-depth.
• Experience building data pipelines and handling sensitive or restricted data.
• Ability to own end-to-end problems in cross-functional settings.
• Comfort working on sensitive projects requiring discretion and integrity.
• Thrive in a fast-paced startup environment with a bias toward action.

Responsibilities

• Design and build secure, sandboxed infrastructure for running sensitive model evaluations, including CBRN and other dangerous-capability domains.
• Build controlled data pipelines and storage for sensitive evaluation material, applying least-privilege and need-to-know access, RBAC, encryption at rest and in transit, audit logging, and data-minimization safeguards.
• Partner with safety researchers to translate evaluation designs into reliable, reproducible, and scalable systems.
• Build eval-orchestration tooling and harnesses for high-throughput evaluations against models in isolated environments.
• Develop infrastructure for measuring AI capability uplift and integrate results into release pipelines.
• Implement guardrails, monitoring, and compartmentalization across compute and data.
• Write production-quality Python and related tooling for data processing and evaluation systems.
• Improve reliability, security posture, and developer experience of the safety platform.

Job description

Reflection is seeking a Research Software Engineer on the Safety team to design, build, and own infrastructure for evaluating open models in high-consequence domains. You will create sandboxed environments, secure data pipelines, and robust tooling to support trusted experimentation.

You will collaborate with safety researchers, legal, and domain experts, translating evaluation needs into scalable, auditable systems with strong access controls and encryption at rest and in transit.