About bylaw
bylaw is an AI safety applied-research lab committed to ensuring AI agents follow rules in the real world. We build evaluations that find legal failure modes, training environments that reduce them, and run-time harnesses that guard against violations in production. Our team is a small, rapidly growing group of engineers, legal experts, and researchers, working together toward safe and law-following AI.
About the role
bylaw is seeking a founding engineer to help translate our frontier research into world-class safety products. This is a high-agency, hands-on role, with the opportunity to shape both the product and the engineering culture we build around it. You should enjoy working with cutting-edge research and turning it into products held to the highest engineering standards in production.
Responsibilities
- Build and scale bylaw's evaluation and training infrastructure, from design docs and architecture through implementation.
- Ship safety tooling to design partners and own it in production.
- Mentor other engineers and raise the technical bar as we grow.
- Contribute to bylaw's overall technical strategy, and proactively identify and address systemic technical and security risks.
Requirements
- 3+ years of software engineering experience.
- Experience building or deploying AI or LLM-based products.
- Experience building customer-facing products or major features from 0-1.
Who you are
- A fast builder who gets stuff done and understands the rigor and demands of a startup.
- A problem solver who is excited about tackling difficult challenges with no predefined solution.
- An owner who takes initiative, works with customers, and builds solutions end-to-end.
Nice to have
- Experience with agentic systems, RAG, or inference infrastructure.
- Experience with VPC, on-prem, or self-hosted deployments for regulated clients.
- Experience building eval harnesses, RL environments, or training/post-training pipelines.
Our stack
- Languages: Python and TypeScript (we're not strict about this)
- Infrastructure: AWS, Kubernetes, Docker
- Databases: Postgres
- Tools: feature flagging, Prometheus, Grafana, Datadog, Claude Code
Join us
We are tackling a difficult problem, but one that, if solved, will enable humanity to harness the benefits of AI while staying in control. We are a team of super-learners, believers, and doers, and we're looking for like-minded people to join our mission: people who ask questions, explore uncertainty, and get hard things done.
Apply
If you are interested in making AI safe in the real world, we would like to hear from you.