Applied Ai Engineer — Customer Intelligence

Build AI that understands an account and knows what to do next
A useful sales recommendation requires more than a company summary. It requires understanding what a customer sells, where they win, what has happened in an account, who matters and why new information should change the next action.Revic combines external company and people intelligence with customer context and account history to help revenue teams identify opportunities and execute. We’re hiring an Applied AI Engineer to build the reasoning, retrieval and evaluation systems that make this reliable.This is a production engineering role, not a prompt-writing or model-demo role.The engineering work
Build customer-specific context. Turn CRM records, conversations, business documents and external research into structured context agents can use. Handle conflicting evidence, missing information and changes over time.Retrieve the evidence that matters. Build and evaluate retrieval and ranking approaches for company, people and interaction data. Preserve sources and distinguish supported conclusions from uncertain ones.Build agents that complete useful work. Develop tool use, structured outputs and multi-step workflows for research, account assessment, signal interpretation and action generation. Decide when a simple model call is enough and when multiple tools and planning are needed.Evaluate quality systematically. Create representative datasets and regression tests. Measure accuracy, relevance, instruction adherence and task completion. Investigate failures across customers rather than optimizing for a few impressive examples.Improve latency and cost without weakening the output. Compare models, retrieval strategies, context selection, batching, caching and fallback behavior. Use measurements to decide what to change.Ship and operate the system. Build the services around the models, instrument agent runs and use production traces and customer feedback to improve performance.What you should bring
Strong Python skills and experience shipping software that uses language models in production.Practical experience with retrieval, tool calling, structured generation or agent workflows.The ability to build evaluations and distinguish model problems from retrieval, data or orchestration problems.Good judgment with noisy, incomplete and contradictory information.Experience making tradeoffs between accuracy, latency, cost and complexity.The ability to debug a multi-step workflow and explain why it produced the wrong result.Product judgment: you care whether the output helps someone complete their job, not just whether it sounds convincing.Experience with search, ranking, entity resolution, information extraction or customer-specific personalization is especially useful. Training foundation models is not the focus of this role.Our stack
Python, FastAPI, the OpenAI Agents SDK, multiple model providers, PostgreSQL, Databricks and Langfuse. We choose tools based on measured performance, not allegiance to one model or framework.Why Revic
You’ll build AI that people use to decide where to spend their time and how to engage a prospect. The challenge is making it accurate, specific, fast and useful across different customers, and you’ll have direct responsibility for how well it works.
#J-18808-Ljbffr