Senior Software Engineer II - Agentic Intelligence

Honeycomb.io

7d•$183,340 - $206,000•Remote

About The Position

Honeycomb is seeking a Senior Software Engineer II with expertise in Agentic Intelligence to join their team. This role focuses on developing AI agents that can investigate, reason, and act on real observability data within Canvas, Honeycomb's agentic workspace. The ideal candidate will leverage deep agent expertise to expand the team's capabilities by building new agents, enhancing Canvas features, and improving performance. The position involves taking code from prototype to production-grade and capitalizing on Honeycomb's unique data store for advanced agent functionalities. This role offers the opportunity to work on cutting-edge AI agent technology in the observability space, contributing to a product used by well-known companies. Honeycomb is a fast-growing, fully distributed company with a strong culture and a commitment to employee growth and well-being.

Requirements

AI and agent engineering experience, with a track record of shipping LLM-based systems relied upon in production.
Experience in designing agent systems and understanding their failure points.
Ability to take ownership of projects from rough prototype to production-grade without extensive oversight.
Current judgment and informed opinions on recent shifts in agent design.
Deep understanding of agent architecture and how fast, high-cardinality data stores impact agent reasoning.
Product judgment to identify and prototype future agent capabilities.
Ability to verify identity and eligibility to work.

Nice To Haves

Observability or developer-tools background.
Familiarity with eval frameworks, agent tooling, RAG, and prompt engineering.

Responsibilities

Design and deliver production-grade agents that investigate, reason, and act on live observability data within Canvas.
Ensure agents are trustworthy for engineers in high-pressure situations, including during incidents.
Take agent projects from initial prototype to a production-ready state.
Own the entire lifecycle of agent work, including scoping, building, shipping, and maintaining agents and their evaluations.
Perform full-stack development as needed to support agent work.
Utilize Honeycomb's high-cardinality data store to build agents capable of advanced reasoning and correlation across services.
Ship new capabilities into Canvas, the MCP server, and Canvas Skills, including memory, spatial awareness, and performance improvements.
Make data-driven cases for future development with working code prototypes.
Define and set standards for agent quality, including measurability through real evaluations, maintainability, and honesty about limitations.