Staff AI Safety Engineer — Build Reusable Safety Tools

Lila Sciences, Inc.; Cambridge, San Francisco (MA, CA); On-site; USD 192,000 - 272,000; Full time

Benefits offered by this job

Medical benefits

Dental benefits

Vision benefits

Life & Disability insurance

Paid time off

Parental leave

Education assistance

Commuter benefits

Lunch program

Equity

Job summary

Lila Sciences, Inc. in Cambridge, MA is seeking a backend software engineer to help build the AI safety toolkit used by researchers and engineers across the organization.

You will own engineering from scoping to deployment, delivering APIs, data pipelines and evaluation tooling that supports threat modeling, red teaming and mitigation workflows. This role emphasizes reliability, performance, and clear communication within a rapidly evolving AI research environment.

Qualifications

• Production software delivery with evolving requirements.
• Experience turning researchers' needs into usable tools.
• Strong Python and backend engineering skills.
• Reliability, performance, and debugging focus.
• Experience in AI research or safety related to evolving workflows.
• Clear communication and interest in scientific AI safety.

Responsibilities

• Build reusable services and tool integrations for evaluations and red-teaming.
• Develop tools to organize evidence, inspect results and review data.
• Turn research needs into maintainable software with cross-team collaboration.
• Ensure reliability and usability through testing and monitoring.
• Collaborate with core AI and software teams to align tooling.

Skills

Python

APIs design

Data pipelines

Databases

Backend engineering

AI safety experience

Research translation

Communication

Tools

Distributed systems

Model-serving integrations

Job description

Lila Sciences, Inc. in Cambridge, MA is seeking a backend software engineer to help build the AI safety toolkit used by researchers and engineers across the organization.

You will own engineering from scoping to deployment, delivering APIs, data pipelines and evaluation tooling that supports threat modeling, red teaming and mitigation workflows. This role emphasizes reliability, performance, and clear communication within a rapidly evolving AI research environment.