Note: The job is a remote job and is open to candidates in USA. FetchData is building intelligence for recruiting through agents that understand context, retrieve information, use tools reliably, and advance work. The Senior AI Engineer will own search, memory, context assembly, agent orchestration, evaluation, and developer tooling for production recruiting workflows.
Responsibilities
- Agent harnesses and orchestration. Design how agents reason, select tools, maintain state, recover from failures, and resume work across conversations and background workflows. Make deliberate decisions about model selection, execution limits, retries, and human handoffs
- Search and retrieval. Improve how Fetch finds and matches clinicians to opportunities. Combine structured criteria, unstructured information, and conversation history; experiment with query planning, semantic retrieval, and reranking to improve relevance
- Tools that accomplish real work. Give agents well-defined capabilities to search records, read CVs, collect missing information, schedule follow-ups, and interact with recruiting systems. Make those actions permission-aware, observable, and safe to retry
- Context and memory. Help agents use clinician preferences, relationship history, job details, and recruiter instructions effectively. Handle incomplete or conflicting information without losing track of what is known and where it came from
- Evals and experimentation. Build repeatable evaluations from real recruiting scenarios and production failures. Measure retrieval relevance, factual grounding, tool use, conversation quality, task completion, and appropriate handoffs alongside latency and cost
- Debugging and developer tooling. Build traces, replay tools, and experiment workflows that make agent behavior understandable and improvements easier to ship. Turn customer feedback and production issues into reproducible cases and lasting fixes
Skills
- Experience shipping and operating production software, with strong backend engineering fundamentals
- Hands-on experience building LLM applications or agents that use tools, retrieve information, or complete workflows
- Familiarity with search or retrieval systems, including the tradeoffs between structured filtering, keyword search, embeddings, and ranking
- Strong judgment about where to use model reasoning, deterministic code, or human input
- Comfort debugging across model behavior, prompts, application code, external APIs, queues, and persistent state
- The ability to take an ambiguous customer problem through implementation, rollout, and iteration
- Proficiency in TypeScript or another backend language, and willingness to work in a primarily TypeScript codebase
- Agents that maintain context and complete tasks across multiple interactions
- Evaluation datasets, simulated conversations, human review workflows, or model-based graders calibrated against human judgment
- Search relevance tuning, recommendation systems, or retrieval over structured and unstructured data
- Agent memory architectures
- Durable workflows, asynchronous systems, and integrations with external business software
- Worked on agents for domains where accuracy and relationship context matter
Company Overview