Langfuse - Senior Backend Engineer

Why Backend Engineering at Langfuse

Your work will have an outsized impact.

Langfuse ingests terabytes of AI tracing data per day. When a Fortune 50 company instruments their application with Langfuse, they hit the infrastructure you built. Everything you ship is open source and immediately visible.

The core of your role is making Langfuse fast and affordable at scale: owning the ingestion pipeline, optimizing ClickHouse data models and queries so dashboards load instantly, and building the backend abstractions that let product engineers ship data-heavy features without reinventing the plumbing each time.

You will also make sure Langfuse runs everywhere. You’ll operate Langfuse Cloud across multiple production environments, and make self-hosting effortless. You will support a single Docker Compose setup to enterprise-grade Helm chart deployment.

You won’t be solving these problems alone. Langfuse is now part of ClickHouse, which means the people who built the database you’re optimizing are one channel away. Few backend roles give you that kind of access to the internals of your core datastore.

You will grow at Langfuse by

• Own and optimize our ingestion pipeline: Langfuse ingests massive volumes of tracing data through a pipeline that flows from our API to ClickHouse. You'll work on throughput, latency, reliability, and cost efficiency at every stage of this pipeline.
• Make ClickHouse fast: our tracing data lives in ClickHouse. You'll design schemas, optimize merge strategies, tune batch inserts, write and review analytical queries that power dashboards and the UI. You'll become an expert on ClickHouse internals — and you'll have the ClickHouse database engineering team one channel away.
• Build new infrastructure-heavy features: things like LLM-as-a-Judge (where we post-process large volumes of ingested data asynchronously), alerting at scale, client-code execution, or batch dataset operations. Build the foundation and, if you like, dip your toes into the frontend for those.
• Test with real data in production: when evaluating a new ClickHouse schema, migration, or query optimization, you'll run load tests against production data to validate that changes actually improve things. We don't guess at performance. We prototype new table designs, run them in parallel on real traffic, and compare before committing.
• Scale for what's next: we're growing fast and new product directions like complex long-running agent observability and real-time evaluation push the infrastructure in new ways. You'll be thinking ahead about what breaks at 10x scale.

Learn more about our infrastructure

• From Zero to Scale: Langfuse's Infrastructure Evolution — deep dive into how we went from Postgres to ClickHouse
• Max talking infrastructure at ClickHouse OpenHouse Amsterdam — recording of our CTO walking through the architecture
• Architecture