Member of Technical Staff Research Engineer
ComposioVisit Composio website
Composio provides AI-agent infrastructure for secure access to and actions across third-party applications.
San Francisco, United States
Funding history
About Composio
Composio is an active AI developer-tools company offering managed authentication, tool access and execution, integrations, and agent runtime infrastructure for AI agents.
Skills
Candidate Availability
Required and preferred rules are kept separate and reflect the wording in the original posting.
About the Role
You will develop large-scale evaluations using real tool-calling data, research semantic tool and execution-path search, and train agentic harness systems. You will work with supervised fine-tuning and reinforcement learning on agent traces, harnesses, and application sandboxes.
Requirements
- Research
- Experimentation
- Language model
- Tool calling
- Evaluation
- Semantic search
- Agent
- Supervised fine-tuning
- Reinforcement learning
- Technical writing
Responsibilities
- Build large evaluations using real tool-calling data
- Measure model performance in long-horizon tool execution
- Research semantic search for similar tools and cached execution paths
- Train agentic harness systems using real tool calls
- Perform supervised fine-tuning on agentic traces
- Train reinforcement learning models using agentic harnesses and app sandboxes
- Prototype and test research experiments quickly
- Take research ideas to production with product and engineering