SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Taskrabbit is a marketplace platform connecting people with Taskers for everyday home services like furniture assembly, handyman work, and moving help. This hybrid role (2 days/week in-office at 130 Sutter Street, SF) reports to the Director of AI Strategy and Enablement.
You'll embed with internal business functions—Marketing, Customer Support, Finance, Operations—to discover, scope, and build Claude-powered agents, automations, and internal tools that solve real workflow problems. This is internal-facing work with no external customer product responsibility. You own deployments end-to-end: from technical discovery and scoping through design, implementation, testing, and production deployment. You'll work solo on most builds, occasionally pairing with subject-matter experts from the business function.
Key responsibilities include:
- Running structured technical discovery with business teams to turn ambiguous ideas into scoped builds
- Independently designing, writing, testing, and deploying Claude-powered agents and automations using Claude Code, Claude API/Agent SDK, or agentic frameworks
- Making and defending real engineering decisions: model selection, context/caching strategy, evals, human-in-the-loop checkpoints, error handling, cost forecasting
- Establishing baselines and defining metrics that prove impact (time saved, error rate, throughput) before and after each build
- Training non-technical teams to identify and build lightweight automations, reducing dependency on this role
- Translating between business problems and technical tradeoffs to move stakeholders from skepticism to adoption
- Scaling successful patterns into reusable playbooks for other teams
Required qualifications: 2+ years hands-on professional experience building AI agents, automations, or internal tools with direct code-level work in Claude Code, Claude API, Claude Agent SDK, or agentic frameworks built on Claude. You must have shipped production systems and be able to discuss architecture decisions and production failures. Strong software engineering fundamentals required—comfortable writing and owning real code (not just configuring no-code platforms). Fluency in API integration, MCP, version control, testing, and debugging. Demonstrated technical judgment on model selection, caching/cost strategy, reliability, and human-in-the-loop design. Track record of embedding with non-technical stakeholders and translating between business and technical contexts.
