Software Engineer, Applied AI
Mantle
- San Francisco, United States
- On-site
- Mid Level
- 2+ years
- Posted
Skills
Required
- AI
- retrieval
- Python / Node.js / APIs / data pipelines
- LLM-powered
- prompt design
- tool calling
- agent orchestration patterns
- AWS
- product development
- AI development tools
- Cursor
- Claude Code
This position is core to our product. Mantle is an AI agent, so the quality, reliability and trustworthiness of that agent is the product. As a Software Engineer on our Applied AI team, you'll design and build the agent systems that take real work off of families' plates: understanding their situation, taking action on their behalf, and knowing when a human needs to step in.
About Mantle
About Mantle
We’re building an AI agent to help families support their aging parents ( origin story here
Over 40M “sandwich generation” Americans currently manage their parent’s affairs. Keeping mom/dad safe and supported is like a second job: it’s time-consuming, stressful and on top of their already busy lives. We believe that managing our parents’ affairs should feel under control, easier and lower stress.
Co-founded by serial entrepreneur partners including Ted Zagat, who sold his previous businesses to Slack ( announcement here) and Google (Zagat restaurant guides).
Responsibilities
- You'll work directly with founders and the engineering team to take agent capabilities from prototype to production. It's not just about clever prompts; it's about the unglamorous work of building evals, chasing down failure cases, and making the agent dependable enough that families can rely on it every day.
- Agent orchestration: The core systems that let Mantle's agent plan, use tools, remember context and complete multi-step tasks for families
- Evaluation frameworks that measure quality, accuracy and safety, and catch regressions before our users do
- Guardrails and escalation paths that keep the agent trustworthy with sensitive personal and financial information
- Retrieval and memory systems that give the agent the right context about each family at the right time
- Integration layers connecting AI models, third-party services and internal systems into cohesive user experiences
- Feedback loops and data pipelines that make the agent better every week
- You love creating things from scratch and taking full ownership
- You measure quality instead of assuming it
Requirements
2-10 years building production software with modern technologies
- Hands-on experience shipping LLM-powered features or agents to real users
- Strong grasp of prompt design, retrieval, tool calling and agent orchestration patterns
- Experience building evaluation systems and using them to drive product decisions
- Strong backend skills (e.g. Python or Node.js, APIs, data pipelines) and experience deploying on AWS
- Ideal Background
Early-stage startup experience (especially as a first-10 engineer)
- Experience building multi-step agents that take real actions on a user's behalf
- Experience designing AI systems that handle sensitive personal or financial information
- Understanding of product development and what makes AI feel trustworthy to everyday consumers
- Nice to have, but not required: experience with FinTech
Who Thrives Here
- You think about user impact, not just technical elegance
- You stay on the frontier of models, tools and techniques, and use AI development tools (Cursor, Claude Code, etc.) every day
- You balance perfection with speed and know when to ship
Preferred
- You thrive with minimal direction and maximum responsibility
Benefits
You care about building something meaningful
- Meaningful work: Build an AI agent that will improve the lives of millions of families, with real technical ownership, fast pace, and huge financial potential
- Small team impact: Work directly with founders and see immediate impact from your work
- Early team equity: Meaningful ownership in a company targeting a massive market
- Shape the future: You will own the core of our product; your decisions will shape how Mantle's agent thinks and acts
- Learning & growth: Work on some of the hardest open problems in applied AI
- Remote Fridays: Work from home every Friday. Monday through Thursday, we build together in person in San Francisco
