Software Engineer I - LLM Observability
Software Engineer I - LLM Observability is an open role at Honeycomb.io, a software company. The role is remote, and the posting gives United States as the location. Our labels mark it as a mid-level role in data and machine learning. Technologies named in the posting: LLM, React, OpenTelemetry, Golang, Python and TypeScript. It asks for 3+ years of experience. Visa sponsorship is not offered. Perks named in the posting: equity and unlimited PTO. Honeycomb.io has 10 open positions on this site, 5 of them in IT and engineering.
Posted 5 days ago · Last checked 2026-10-21
About Honeycomb.io
Software product company 10 open positions, 50% in IT and engineering. Most open roles: Data and machine learning (4), Software engineering (general) (4), Non-software engineering and technicians (2).
Similar jobs
1,052 open data and machine learning jobs in the United States · 727 open remote data and machine learning jobs
- AI Engineer — Ignite Digital, San Diego, California, United States
- Senior Privacy Review Specialist (AI/LLM Governance) — Exadel Inc (Website), United States of America
- Applied AI Engineer — ElevenLabs, United Kingdom | Dublin | Spain | Hungary | Bulgaria | Germany | Poland | Italy | Netherlands | France
- Data Scientist / Visualization Specialist - SCON — Accenture Federal Services, Washington, DC
- AI and Data Engineer — Accenture Federal Services, Tampa, FL
- Principal Data Scientist, User Ecosystem — Reddit, Remote - United States
- Senior Data Scientist — Oscar Health, New York, New York, United States
- Senior ML Engineer, Perception — Clera, San Francisco
| Workplace | Remote |
| Seniority | Mid |
| Role | Data and machine learning |
| Countries | United States |
| Languages | GolangPythonTypeScript |
| Key facts | Stack: Golang, Python, TypeScript, LLM, React, OpenTelemetry · Experience: 3+ years · Remote · Visa sponsorship: no · Schedule: on-call · Equity · Unlimited PTO |
The full description is on the original posting. Labels are set automatically and can be wrong. About the data