About Sonar

Sonar is a mission-driven startup founded out of Stanford that is redefining society's ability to deliver personalized mental health support at scale . Sonar provides 24/7 AI-enabled, human led, chat based support with trained coaches. This unique human-in-the-loop approach allows us to deliver safe, accessible and effective support at a reasonable cost and has been recognized by the New York Times, Wall Street Journal and CNN.

Since our founding in 2022, we have scaled to support more than 20,000 youth across the United States, delivering improved mental health, grades, attendance and reduced tragedies. We are backed by leading institutions such as Stanford University, Informed Ventures, HopeLab and Nina Capital. We have 10 full-time employees, 4 part-time employees and 12 wellbeing coaches, all working remotely.

Role Overview

In this role, you’ll help design, test, and improve AI systems that power key parts of our product and internal operations. You’ll work on projects such as developing evaluation frameworks, improving agent performance, and creating tooling that helps our AI systems learn and improve over time. The role blends experimentation, analysis, and hands-on building in a fast-moving environment.

What you’ll be doing

• Design, build, and improve agent-based AI systems for internal workflows
• Develop internal tooling and infrastructure for evaluating AI system quality and performance
• Build frameworks to benchmark models, vendors, and system configurations to inform product decisions
• Improve the quality, reliability, and effectiveness of agents through experimentation and testing
• Partner cross-functionally to translate product and operational needs into automated processes.
• Contribute to the design of memory, feedback, and workflow systems that improve AI performance over time
• Contribute to our codebase to implement and improve AI systems and tools

Who you are

• AI Enthusiast: Curious about how AI can be used in meaningful ways to augment humans, and regularly experiments with new models and tools to understand their potential applications.
• Systems-Thinker: Able to connect the dots between models, prompts, data, tools, and product workflows, ensuring AI systems work effectively within the broader platform.
• Data-Driven: Comfortable working with datasets, SQL queries, and unstructured data to analyze AI system performance and inform improvements.
• Independent: Able to take loosely defined ideas or problems and turn them into working implementations.
• Research-Minded: Excited about leveraging the latest advancements in research to build AI systems that are useful, reliable, and continuously improving.

Qualifications

• Strong proficiency in Python and SQL
• Familiarity with GitHub and modern software engineering workflows
• Experience building projects involving agents, LLM-based systems, or related AI applications
• Clear evidence of shipping and implementing technical projects, whether in industry, research, or high-quality side projects
• Experience working with conversational datasets

Nice-to-have

• Bachelor’s or Master’s degree in computer science, data science, or a related field
• Experience working with mental health and/or healthcare data
• Familiarity with LangGraph and Pydantic for multi-agent orchestration
• Experience designing or running AI evaluations
• Contributions to open-source AI repositories

Other Stuff

• Cash: $90-120K per year, depending on experience and credentials
• Equity: Meaningful equity, commensurate with experience
• Location: Fully remote (US-based)
• Unlimited PTO
• Health benefits