Founding Engineer, Agentic Systems

Champ AI•San Francisco, CA

About The Position

Champ AI is building a multimodal work-agent orchestration platform that helps operations, support, and compliance teams automate end-to-end workflows. The Founding Engineer (Agentic Systems) will own core parts of the agent runtime and developer and product surface area, building systems that enable agents to operate safely, deterministically, and measurably in production. This hands-on role involves shipping quickly, working directly with design partners, and helping define standards for enterprise-grade agents. The engineer may own areas including agent runtime and orchestration, context and memory management, multi-agent patterns, tool frameworks and integrations, data syncing, browser automation, secure sandbox execution, evaluations and reliability, experimentation and human review loops, and AI-native product experiences. The role has high ownership, fast iteration, direct customer feedback, and meaningful influence on architecture, product direction, and hiring.

Requirements

  • Has shipped AI agents into production, internally or externally, or made meaningful open-source contributions in agent frameworks, evaluation tooling, RAG/memory tooling, browser automation, or similar areas.
  • Experience with LLM tool use, structured outputs, function calling, and multi-step workflows.
  • Experience with context engineering, including retrieval strategies, chunking, reranking, summarization, and memory read/write policies.
  • Systems-thinking experience with state machines, retries, idempotency, failure modes, and operational reliability.
  • Experience with integrations, including OAuth, scopes, token refresh, pagination, incremental sync, webhooks, and rate limiting.
  • Experience with sandboxed execution or secure-by-default infrastructure patterns, such as containers, ephemeral environments, and secrets management.
  • Experience with observability, including traces, metrics, and logs, and building explainable runs for people.
  • Experience evaluating non-deterministic systems, including confidence scoring and regression testing.

Nice To Haves

  • Experience building AI-native user interfaces, such as agent run views, trace explorers, or approval interfaces.
  • Experience with enterprise requirements such as SOC 2 posture, auditability, access controls, and tenant isolation.
  • Ability to move between research-oriented prototyping and production-grade engineering.

Responsibilities

  • Design agent loops for planning, tool use, verification, and recovery with strong guardrails.
  • Build context assembly pipelines involving retrieval, compression, summarization, and state that persists across long workflows.
  • Develop short- and long-term memory systems for users, organizations, and projects, including safe memory write policies.
  • Implement multi-agent patterns such as delegation, handoffs, coordinator/worker setups, and concurrency.
  • Build typed tool-definition frameworks with validation, retries, idempotency, rate limits, and observability.
  • Develop SaaS connectors and data-sync systems using APIs, webhooks, polling, incremental sync, and conflict resolution.
  • Build browser automation and computer-use flows, including authentication, session handling, and handling DOM variability.
  • Create secure execution environments for agent-written code, scripts, and data transformations, including permissions, isolation, secret management, and audit trails.
  • Support deterministic replays, dry-run modes, and blast-radius controls where possible.
  • Build evaluation harnesses for tool-use correctness, workflow completion, policy compliance, and regression detection.
  • Develop golden and synthetic tasks and use production traces to measure offline and online performance.
  • Build experimentation frameworks for prompt, model, and tool changes, including versioning and rollbacks.
  • Create human-in-the-loop review processes for sampling, labeling, adjudication, and continuous improvement.
  • Build AI-native interfaces that make agent behavior understandable and controllable, including traces, state, explanations, and editable plans.
  • Develop UX for approvals, step-through execution, partial automation, and exception handling.
  • Provide customer-facing debugability through audit logs, run history, and data provenance.
  • Ship to production quickly, work directly with design partners, and contribute to architecture, product direction, and hiring.

Benefits

  • Competitive salary
  • Meaningful equity
© 2026 Teal Labs, Inc
Privacy PolicyTerms of Service