Data Scientist (ML + LLM Systems)

Role overview Build and improve machine learning and large language model systems that support real customer-facing marketing decisions. The work spans agent workflows and classical ML, including audience segmentation, lead scoring, churn prediction, journey optimization, content generation, streaming event pipelines, and recommendations. You will help balance model quality, response time, reliability, and inference cost in production.

Responsibilities - Develop and maintain LLM applications, including tool-using agents, retrieval-augmented generation, and multi-model orchestration. - Improve a multi-stage agent pipeline, its tool functions, streaming responses, and safeguards around actions with side effects. - Tune model routing to manage inference costs while maintaining useful output quality. - Build and refine streaming ML pipelines for behavioral scoring, likelihood estimates, and next-best-action signals. - Contribute to recommendation systems that use graph methods and text or image embeddings. - Make deployed systems understandable to business users, including explaining model decisions in accessible terms.

Requirements - At least three years of experience delivering ML systems used by real users. - Practical experience developing LLM applications such as tool-calling agents, RAG pipelines, or multi-model systems. - Strong Python skills and ability to work with Java-based platform services. - Experience with streaming data and real-time feature engineering, such as with Flink or Kafka. - Ability to weigh inference cost alongside model quality and operational needs. - Familiarity with graph-based machine learning and recommendation systems.

Nice to have - Experience with Neo4j, CLIP, or SentenceTransformer embeddings. - Experience with PyTorch or scikit-learn.

Benefits and work setup - Full-time, remote position.