SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Salary: USD 151,300 - 229,900 / annual
Thumbtack is a home services marketplace that helps millions of homeowners find local service professionals and manage home projects. The Business Applications team builds and operates internal systems that keep Thumbtack running at scale, including Salesforce-centered workflows, admin tooling, and integrations across the company. The team is now expanding into a broader internal platform for AI-enabled work, with shared APIs, reusable integration patterns, and safe, auditable capabilities for business-critical actions.
In this role, you will own the product surface that turns agent capability into something business teams can reliably depend on. The core challenge is making AI agents trustworthy: giving teams a way to configure what agents are allowed to do, review and correct their work, and understand what happened and why.
You will design and build interfaces that business teams use to configure agents, review their work, approve or correct actions, and understand agent behavior. You'll build observability and evaluation tooling that makes agent behavior legible—showing where it succeeded, where it failed, and whether changes improved outcomes. Working across the full stack (React, TypeScript, Go, PHP, Python), you'll own features end to end from build through test, launch, and iteration. You'll partner directly with business teams to translate operational problems into working software and own the full lifecycle of your work, including monitoring and post-launch improvement.
REQUIREMENTS:
- 3+ years of professional software engineering experience
- Strong frontend skills with React and TypeScript, including complex application state and data-dense interfaces
- Solid backend engineering ability; experience with Go, PHP, or Python preferred, but strong engineers are expected to pick up new languages
- Working knowledge of relational and non-relational data stores (Postgres, DynamoDB, or similar)
- Real product judgment: ability to evaluate workflows, decide what to build, and identify what to cut
- Interest in problems of non-deterministic systems—uncertainty, partial failure, and designing for human correction
- Proven ability to navigate ambiguity, close knowledge gaps, and deliver in evolving environments
