AI Engineer

As an AI Engineer at CCC Intelligent Solutions Inc., you will build and ship LLM-powered features and agentic workflows for customers. These features automate routine parts of a claim and guide customers through the rest, ensuring trust through measured accuracy—not just a convincing demo. Agents operate as steps within deterministic workflows, with privacy and security as design baselines, not review gates. Agentic coding tools accelerate development, but you own the critical aspects: system design, security, and rigorous debugging. You will deliver features end-to-end.

Key Responsibilities:

• Ship LLM-powered features and agentic workflows in Python and TypeScript, from problem scoping with product managers and data scientists to production traffic. Define and track the metrics for success.
• Develop evals, guardrails, and feedback loops to ensure trustworthy features, adhering to explicit token and latency budgets.
• Leverage agentic coding tools daily, reviewing their output for correctness, security, and design before merging pull requests.
• Maintain and extend NLP pipelines that convert unstructured claim data into structured data, blending semantic search, classical NLP, and LLMs.
• Build and operate durable, event-driven services where agents function as steps in deterministic workflows, with checkpointed model outputs to prevent redundant LLM invocations.
• Trace every model call and agent step to enable reconstructible and auditable recommendations, including citations to source documents, and integrate tracing into alerts.
• Model and query claims data across relational, document, and graph stores (e.g., PostgreSQL, MongoDB, Neo4j).
• Design privacy and security into features from the start, including tenant isolation, least-privilege access for agents, and clear rules for data reaching model providers.
• Write technical design documents before significant changes, using them to gather feedback and buy-in from the team before development.
• Participate in a shared on-call rotation for the services you build.

Requirements:

• 3+ years of professional software engineering experience, including at least one LLM-powered feature shipped and operated in production.
• Proficiency in Python and TypeScript, including typed, tested, and async code in both, with deep expertise in one.
• Experience designing evals for LLM output and shipping guardrails.
• Clear understanding of how to constrain agentic coding tools’ context and verify their output, with daily use of tools like Claude Code, Cursor, or GitHub Copilot on production work.
• Pipelines blending LLMs with semantic search and classical NLP, including structured output, tool calling, embedding-based retrieval, and cases where classical techniques outperformed LLMs.
• Experience building services on message- or event-driven architectures, including long-running workflows with retries, idempotent handlers, and event-sourced state, plus an LLM or agent step within one.
• Data modeling and development across relational, document, and graph databases (e.g., PostgreSQL, MongoDB, Neo4j), including knowledge of when to use each and partition-key, consistency, and tenant-isolation strategies.
• Backend services built and operated on Azure or AWS, with exposure to durable workflow engines like Temporal or Durable Task Scheduler.
• Experience writing technical design documents to gain team feedback and buy-in before development.
• Clear communication skills for both technical and non-technical audiences, along with strong code review abilities.

Even Better if You Have:

• React and TypeScript experience to deliver full-stack features end-to-end, especially for chat, streaming, and citation-heavy AI experiences.
• Experience with CI/CD pipelines and infrastructure as code (e.g., Terraform or Bicep).
• Experience with agent frameworks like LangGraph or Microsoft Agent Framework, including knowledge of protocols like Model Context Protocol (MCP), Agent2Agent (A2A), command-line tools, and microservices.
• Experience with LLM observability and evaluation tooling such as LangSmith, Arize, or Braintrust, and OpenTelemetry-based tracing.
• Experience designing A/B tests for AI features or what-if simulations that replay historical claims through proposed changes before shipping.
• Exposure to the insurance, claims, or automotive repair domains.

What Success Looks Like:

• In your first six months, ship an LLM-powered feature with evals running in CI and clearly defined metrics for success.
• When a model migration or prompt change causes a regression, your evals catch it before a customer experiences issues.
• Take ownership from concept to production with minimal direction, asking the right questions early.
• Ensure your services are resilient: they survive retries, partial failures, and model migrations without customer-facing incidents.
• When something goes wrong in production, the trace identifies the specific step and model call, and you can show an auditor the same information.
• Privacy and security reviews of your features find no issues not already addressed in design.
• Enjoy the craft and demonstrate it: engineers adopt your patterns without explicit instruction.
• Engage in the back-and-forth scoping process and can argue a position without needing to win it.
• Embrace CCC’s values of ambitious, collaborative, and empathetic problem-solving, focusing on measured outcomes over arguments.