Software Engineer, AI Agents\n============================\n\nAbout Reacher\n-------------\n\nWe're the #1 TikTok Shop partner helping brands like Under Armour, Hanes, HeyDude, and Logitech scale their affiliate marketing. We've crossed 7 figures in ARR and are backed byY Combinator, Pioneer Fund, Rebel Fund, and operators including Steve Huffman (CEO of Reddit), Matteo Franceschetti (CEO of Eight Sleep), Theo Browne (t3.gg).\n\nWe're making a big bet on AI agents and what they can do for creator marketing. This role sits at the center of that.\n\nWhat You'll Do\n------------------\n\nAI Agents\n\n* Build and own long-running AI agents that automate complex, multi-step workflows for TikTok Shop brands.\n \n* Design agentic systems end-to-end: orchestration, state management, error recovery, and observability.\n \n\nBackend & Infrastructure\n\n* Own the backend systems your agents run on — design, implement, ship, maintain. Python, FastAPI, PostgreSQL, GCP.\n \n* Contribute to our jobs infrastructure (GKE-based, on-demand workers) as needed to support agent reliability and scale.\n \n* Set the bar for code quality and architecture decisions across the backend.\n \n\nYou're a Fit If\n\n* 3+ years of backend engineering experience, ideally at a product-first company\n \n* You've built automation, agents, or complex async systems — and you think deeply about failure modes\n \n* Strong systems thinking, but applied toward agent reliability and workflow design, not just raw infra\n \n* Comfortable with ambiguity and own problems end-to-end without hand-holding\n \n* You've worked directly with customers or product teams and can translate vague requirements into technical decisions\n \n* Resourceful: you figure things out, you don't wait for the answer to show up\n \n\nTechnical Skills We Value\n\n_AI and Agentic Development:\n\n* AI agent frameworks and LLM orchestration for building and managing long-running, multi-step agents\n \n* Claude Code proficiency: context engineering (CLAUDE.md), custom commands, MCP integrations, subagent coordination\n \n* Strong judgment for when to leverage AI vs. when to code directly\n \n* Ability to decompose complex engineering work into tasks suitable for AI assistance\n \n* Experience verifying AI-generated code for correctness, security, and maintainability\n \n\n_Backend:\n\n* Python, FastAPI, PostgreSQL\n\n* System design with a focus on async, long-running, stateful workflows\n\n* GCP/cloud infrastructure (GKE/KEDA) — enough to own what your agents need\n\n* API design (REST/GraphQL)\n\n\n_General:_\n\n* Testing frameworks (Playwright, Pytest) — especially for validating agent behavior end-to-end\n* Code reviews, documentation, and establishing conventions\n\nBonus Points\n------------\n* You've been an early engineering hire at a startup\n* You've built or maintained internal automation systems that ran at scale\n* You've worked on AI agents, LLM pipelines, or autonomous systems\n* You've shipped side projects or contributed to open source\n* You're curious about e-commerce, creator marketing, or what AI will do to the agency business\n\nWhy Join\n--------\n* Small team (5 engineers), high ownership — you're working on the hardest problems, not feature tickets\n* You'll work closely with the founders in SF on the longest-term bets in the company\n* Real equity, we care about it as much as salary\n\nLocation\n--------\n\nOn-site in San Francisco. This role is meant to be close to the founding team.\n\nWhen Applying\n-------------\n\nSend a short note about something you've built and why this role excites you. GitHub, LinkedIn, resume all welcome, but I care more about how you think and what you've shipped.