Benefits
- Equity
- Health insurance
Skills
Full job description
About the company
Our client is a chemical manufacturing technology company.
The role
Raydar is recruiting for this role on behalf of our client. Design and build the core execution layer of an AI-driven product, covering runtime orchestration, tool integration, state handling and the reliability and evaluation work that lets it run in production. The work spans system architecture through deployment and is shaped by close collaboration with researchers and domain experts.
What you'll do
- Build a runtime that coordinates model calls, tool invocations, retries, checkpoints and handoffs to people across long-running multi-step jobs.
- Structure multi-step processes as graphs linking software agents, external tools and decision points so they can be traced and resumed after a failure.
- Wrap external scientific engines, data stores and laboratory systems behind well-tested interfaces.
- Create evaluation harnesses, tracing and regression suites that measure outcomes at the level of the end user's task.
- Help model scientists build the data pipelines that record model runs and feed results back into training.
- Take responsibility for keeping the live system fast, economical and dependable under heavy load, including monitoring and performance tuning.
- Work with domain experts and product engineers to turn specialist workflows into system capabilities.
- Help set team coding norms and architectural choices, coach colleagues, and take part in public code contributions where suitable.
Requirements
What we're looking for
- 2 to 6 years of relevant professional engineering experience.
- Strong production Python skills across application code and infrastructure.
- Hands-on history creating a system that sequences the steps and state of several AI components working together.
- Practical experience with models invoking external functions and multiple autonomous components cooperating, going beyond simple prompting or search-augmented setups.
- Sound understanding of managing shared data, parallel execution, error scenarios and repeatable operations for lengthy background tasks.
- Prior work on monitoring and testing tools for language-model systems, using quantified results to guide fixes.
- Ability to learn a new scientific domain from subject-matter experts; a chemistry background is not required.
Bonus points
- Background in building data collection pipelines or simulated training environments for reinforcement learning is a plus.
- Prior work connecting technical software libraries, organized datasets or linked-entity stores to automated AI processes is a plus.
- A track record of published research, public code or production systems reflecting independent technical achievement.
- A master's degree or unfinished PhD is welcome; a PhD is also welcome.
Benefits
Compensation and benefits
- Base salary: USD 200,000 to 300,000 per year
- Equity
- Health, dental and medical coverage
Location and work model
- San Francisco, California, United States
- On-site
- Full-time
Opens Credence’s own application page in a new tab.
