Software Engineer, GenAI

About the role

Abridge builds generative AI technology designed to transform medical conversations into structured clinical documentation. We are seeking engineers to join our Builder team to develop LLM-driven workflows that operate within complex clinical environments. You will work alongside clinicians and researchers to deploy reliable, high-uptime systems that prioritize accuracy and clinical trust. This role centers on architecting intelligent systems that bridge the gap between raw clinical dialogue and actionable, structured records. You will be responsible for ensuring these systems function seamlessly in high-stakes, real-world medical settings. The work requires a deep commitment to both engineering excellence and the nuances of healthcare data. You will play a key role in defining how generative AI integrates into the clinical workflow.

Key facts

What you'll do

• Engineer agentic systems that leverage advanced retrieval mechanisms and structured tool integration to handle complex clinical reasoning tasks.
• Partner with infrastructure and machine learning specialists to refine model latency and optimize context window utilization for healthcare workloads.
• Design robust data pipelines, intelligent caching frameworks, and scalable generative output architectures suitable for clinical production environments.
• Construct comprehensive evaluation methodologies, including automated benchmark suites and human-in-the-loop testing protocols, to uncover systemic failure modes and edge cases.
• Oversee the complete production lifecycle of software, implementing monitoring dashboards, observability tooling, safety guardrails, and rigorous version control practices.
• Experiment with novel model capabilities and advanced prompting strategies to systematically enhance system reliability and output quality.
• Translate intricate clinical requirements into technical specifications that guide the development of reliable AI-assisted documentation features.
• Collaborate with cross-functional stakeholders to define success metrics and validate the accuracy of generated clinical notes.
• Implement security and compliance best practices directly into the software development lifecycle to protect sensitive patient information.
• Mentor junior engineers by providing code reviews, technical guidance, and knowledge-sharing sessions focused on generative AI applications.

Requirements

• Bring a minimum of 3 years of hands-on experience in production-grade software engineering with a strong track record of delivering reliable systems.
• Demonstrate at least 1 year of focused experience building products powered by Large Language Models or agentic architectures in a professional setting.
• Show advanced proficiency with LLM APIs and orchestration frameworks such as LangChain or LlamaIndex, along with sophisticated prompting methodologies.
• Possess substantial experience designing and integrating retrieval systems, including vector databases, semantic search, and kNN indexing techniques.
• Exhibit strong Python programming skills, including expertise in asynchronous programming, automated testing frameworks, and deployment tooling for cloud platforms.
• Have a proven history of building evaluation datasets and executing A/B testing strategies to drive measurable improvements in model performance.
• Be prepared to maintain a hybrid work schedule, working from either the San Francisco or New York office a minimum of 3 days each week.
• Hold the legal right to work in the United States without requiring sponsorship for this position at this time.

Nice to have

• Hands-on experience with function calling patterns and the implementation of complex agentic workflows involving multi-step reasoning.

Skills & tools

• Python
• LLM APIs
• LangChain / LlamaIndex
• Vector Databases
• Retrieval Systems
• Automated and Human-in-the-loop Evaluation

Practical notes

Relocation assistance is provided for candidates moving to San Francisco. Comprehensive benefits include health, dental, and vision plans, 401(k) matching, paid parental leave, family forming support, flexible PTO, 14 paid holidays, and a monthly lifestyle wallet. Abridge offers a sabbatical program after 5 years of service. All official communication will originate from an @abridge.com email address. Applicants are encouraged to move quickly as roles are filled on a rolling basis. The standard work week is 40 hours. This position is not eligible for visa sponsorship at this time. The role requires a high degree of collaboration, necessitating consistent in-person presence on specified days.