Responsibilities
- Independently map business workflows and develop AI agents or sequences of agents to automate them
- Iteratively improve agent performance by refining prompts, managing edge cases, and increasing reliability through real-world use
- Rapidly prototype in early stages, then apply rigorous engineering standards when advancing agents toward production
- Construct and manage data pipelines that extract from source systems like Microsoft Graph, Teams, and Snowflake into a data warehouse, applying filtering, summarization, and sensitivity handling before indexing
- Use integration platforms such as Workato to create and maintain workflows and API connections between systems, including monitoring and logging
- Understand how enterprise search and AI indexing tools like Glean process data, including index scope, group-based access limits, and permission-aware retrieval
- Implement proper access control mechanisms such as privileged access zones, IP allowlisting, OAuth-protected endpoints, and group-based data restrictions
- Account for identity and access management systems like Okta/SSO and security monitoring tools like Panther to avoid security blind spots
- Apply data sensitivity tagging and minimization when extracting raw data, removing unnecessary content and redacting employee-specific information
- Promote strong software engineering practices across agent development, including version control, code reviews, testing, and deployment discipline
- Define clear expectations for production readiness distinct from prototype phases, helping the team identify development stage
- Take full ownership of agents from concept through production, including error handling, monitoring, and recovery from failures
- Leverage and extend existing shared systems instead of recreating functionality, following an 'extend, don't rebuild' approach
- Gain practical understanding of sales, customer success, and commercial operations to ensure agents reflect actual workflows
- Collaborate closely with a peer agent developer, sharing design and implementation tasks flexibly
- Integrate engineering best practices without slowing down early-stage experimentation and prototyping
