Agentic Engineer ($160k-$220k + Equity) at Snapp Stats

Job Title Agentic Engineer Salary $160k-$220k Equity Company Description Snapp Stats is an a16z-backed startup building the ultimate AI conversational companion for sports fans and bettors. Led by repeat founders, including a co-founder of Caviar, the team leverages agentic systems and real-time sports data to deliver personalized insights and proactive recommendations to passionate fans and fantasy players. Job Description As an Agentic Engineer at Snapp Stats, you will own the orchestration layer of AI agents designed for sports fans. You will design robust agent loops, manage tool-call latencies, and implement sophisticated evaluation frameworks. This role bridges the gap between experimental LLM hacks and production-ready systems that thrive under real-world user interaction. Location San Francisco, USA Why this role is remarkable Work at the cutting edge of agentic AI with a16z backing and a leadership team featuring the co-founder of Caviar. Direct impact on a product where users notice accuracy and speed immediately, operating at the emotional intersection of sports and betting. High-autonomy environment where your architectural decisions and tool designs go from concept to production within the same week. What You Will Do Design and ship complex agent loops that manage long-horizon tasks, tool calls, and partial failures without degradation. Build and operate MCP servers to expose rich sports databases and content tools to both internal and partner agents. Establish rigorous evaluation harnesses and observability pipelines using tools like Logfire to monitor token spend, latency, and reliability. The ideal candidate 5 years of software engineering experience with at least 1.5 years shipping and operating production agentic systems for real users. Deep proficiency in production Python, async debugging, and agent frameworks like Google ADK, LangGraph, or Pydantic AI. Strong discipline in evaluations and observability, with the ability to move beyond vibes-based prompting to data-driven regression testing. #J-18808-Ljbffr Jack & Jill