About the job
About Blee
At Blee, we are revolutionizing traditional workflows that often rely on spreadsheets and emails by developing AI-first software that integrates the entire organization to meet compliance requirements swiftly, comprehensively, and transparently. Currently, we support thousands of active users across numerous enterprise-grade clients with pledged or paid long-term contracts. Our leadership team comprises former professionals from renowned companies such as Adobe, Chime, PayPal, AWS, Wachtell, Harvard Law, and Columbia Law.
We have created the first platform specifically designed to tackle the distinct challenges our clients encounter from start to finish. The ideal candidates for Blee are not only adept problem solvers but are also excited about uncovering the specific issues our users need addressed, many of whom are experiencing a product like Blee for the first time.
Our headquarters is located in downtown San Francisco, with a new office opening in NYC soon, and our team is primarily distributed between the Bay Area and New York City.
The Role
As an AI Engineer specializing in agent system design, you will architect and implement cutting-edge intelligent compliance systems. Your responsibilities will include designing frameworks where multiple agents, powered by both custom and foundational models, collaborate to identify, monitor, and elucidate compliance flags. Additionally, you will construct and enhance applied AI/ML pipelines, fine-tune models, develop evaluation frameworks, tailor AI configurations for customers, and train models based on user feedback to deliver more precise compliance recommendations.
What You'll Do
Design Multi-Agent Systems – Develop and implement frameworks that combine foundational LLMs and custom-trained models to drive intelligent compliance workflows.
Optimize AI/ML Pipelines – Focus on model fine-tuning, prompt engineering, and establishing evaluation frameworks.
Ensure Explainability and Traceability – Create systems for transparent flagging, ensuring that each compliance flag can be traced back to previous rules, examples, or data assets.
Automate Compliance Rule Translation – Develop platforms that convert compliance rules into agent-ready logic paths.
Train and Fine-Tune Models – Enhance compliance accuracy and adaptability by leveraging user feedback and outcomes data for iterative training and personalization.
