Applied Scientist

Blee•San Francisco, CA

20h

About The Position

Blee is hiring an Applied Scientist specializing in agent system design to architect and implement intelligent compliance systems. The role involves designing frameworks in which multiple agents, powered by foundational and custom-trained models, work together to surface, track, and explain compliance flags. You will build and optimize applied AI/ML pipelines, work on fine-tuning, prompt engineering, evaluation frameworks, and customer-specific AI configurations, and train models using user feedback to produce more targeted compliance recommendations. You will collaborate closely with Blee's AI and engineering leadership, product engineers, implementation managers, model training engineers, and compliance users. The role is based in the Bay Area or New York, with an expectation of coming into an office 2–3 days a week; remote work may be possible for the right person. Blee describes its environment as high-trust and high-autonomy, with an emphasis on clarity, ownership, shared responsibility, and in-person collaboration.

Requirements

At least 5 years of applied ML/AI engineering experience.
Strong knowledge of agent-based architectures and orchestration frameworks.
Strong Python skills and proficiency with modern software engineering practices, including APIs, microservices, and distributed systems.
Experience with ML frameworks such as PyTorch, TensorFlow, and Hugging Face.
LLM experience including fine-tuning, prompt engineering, RAG, and evaluation.
Background in data preprocessing, feature engineering, and training pipelines.
Production ML experience, including model serving, monitoring, A/B testing, and data pipelines.
Strong communication skills and ability to work cross-functionally.
Based in the Bay Area or New York and willing to come into the office 2–3 days a week.

Responsibilities

Architect and implement multi-agent systems that integrate foundational LLMs and custom-trained models for compliance workflows.
Build and optimize AI/ML pipelines, including model fine-tuning, prompt engineering, and evaluation frameworks.
Develop mechanisms for transparent flagging so each compliance flag can be traced to prior rules, examples, or data assets.
Build systems that translate compliance rules into agent-ready logic paths.
Design, train, and fine-tune models using user feedback and outcomes data to improve compliance accuracy and personalization.
Design model evaluation and monitoring infrastructure to support reliability and auditability across customer deployments.
Collaborate with product engineers to integrate AI features through APIs, data pipelines, and frontend systems.
Partner with model training engineers to incorporate feedback loops into system design.
Work with compliance users to translate complex rules into agent workflows.
Prototype and apply new AI/ML techniques to the platform.