Data Engineer

Azumo builds and operates production AI systems for companies ranging from seed-stage startups to Meta. We are hiring a Data Engineer to own the layer everything else depends on: ingestion and transformation pipelines, storage and warehouse design, and the retrieval infrastructure that AI systems query. The role is fully remote across Latin America, aligned to your client's working day.

You will not be building pipelines that work until the schema changes. Azumo has shipped production data systems since 2016, and the work here is judged downstream: whether a model can be trained on what you deliver, whether a retrieval query returns the right passage, whether a number survives being questioned by the client.

Where this role sits

Azumo's engineering organization is built around four lanes. The Data Scientist lane owns the question and the method. The AI Engineer lane owns production behavior. The Software Engineer lane owns AI-augmented product delivery. This role is the Data Engineer lane, and it owns pipelines, storage, and the retrieval layer.

One question places the boundary: when the output is wrong, whose problem is it? "The data was missing, stale, or wrong by the time it arrived" is yours."The system did the wrong thing with data that was correct" is the AI Engineer's.

Not quite your profile? Check our other openings:

• If you decide what to measure and which method answers it — Data Scientist

• If you own how an AI system behaves in production — AI Engineer

• If you ship product software with agents in your toolchain — AI-Augmented Software Engineer

• If you've done all of the above and answered to the client directly — Forward Deployed Engineer

What you will build

• Ingestion and transformation pipelines. Batch and streaming ingestion on Spark, Kafka, dbt and Airflow, with idempotency, backfills, schema evolution and late-arriving data handled by design rather than by hand.

• Storage and modeling. Warehouse and lakehouse design on Snowflake, BigQuery, Redshift or Databricks, with partitioning, file layout and query cost treated as engineering decisions.

• The retrieval layer. The chunking, embedding and indexing pipelines that feed RAG systems on pgvector, Pinecone, Qdrant or Azure AI Search, and the freshness, deduplication and permission problems that come with them.

• Data quality as a contract. Tests, expectations, lineage and alerting. If a pipeline is wrong, the people downstream should hear it from you and not from the client.

• Sensitive data by default. PII classification, masking, row and column level access, retention and deletion, and an audit trail that holds up when a client asks who read what.