Software Engineer, AI Agents
In-Office
Posted 10d ago
Remote Work Policy
On-site
Categories
AI Agent Engineer
About the job
At Cloudflare, we are building industrial-scale AI agents that directly support customers, moving beyond research to power real customer interactions globally from day one. This role involves assembling existing Cloudflare technologies like Workers, Durable Objects, KV, R2, D1, Vectorize, Workers AI, AI Gateway, and the Agent SDK into functional agents. The primary goal is to ship production agents on the Cloudflare stack, iterating quickly through building, deploying, and learning. Your code will be the primary interface for Cloudflare customers, directly impacting their experience.
Responsibilities
- Build agents on Workers using Durable Objects for state and memory.
- Integrate tools using the Agent SDK, MCP, and function calling.
- Utilize Vectorize, KV, R2, and D1 for memory, cache, files, and configuration.
- Run models through Workers AI and AI Gateway, integrating third-party models as needed.
- Create and implement evaluation frameworks, guardrails, and audits for rapid tuning and iteration.
- Develop agents capable of summarizing information, proposing solutions, and escalating issues to humans.
- Expose agent health and metrics through transparent dashboards.
- Integrate with queues and webhooks, publishing events on Queues or Pub/Sub.
- Reduce cost per case and time to first response, validated by data.
- Take end-to-end ownership, including on-call responsibilities with team support.
- Design and maintain robust observability for distributed AI workflows, including structured logging and end-to-end tracing.
- Architect security boundaries for agent operations, implementing secure credential handling and multi-layer approval gates.
Requirements
- Demonstrated success in shipping production systems with tangible work samples.
- Proficiency in TypeScript or Rust for Workers development, including HTTP, queues, async operations, and performance optimization.
- Hands-on experience with Durable Objects, KV or R2, and D1 or Postgres.
- Experience with model tooling, including prompt I/O, tool calling, evaluations, and safety checks.
- Strong understanding of observability principles (logs, traces, metrics).
- Experience with agent-to-agent (a2a) or multi-agent frameworks.
- Experience developing LLM evaluation frameworks, including automated scoring and CI-integrated quality gates.
- A bias for simple, scalable design principles.
Benefits
- Impact at global scale
- Work on the edge platform
- Career growth and leadership opportunities
- Culture of ownership, autonomy, and accountability
- Opportunities for learning and collaboration across teams