Data Engineer

Frontier's AI runs on data — training signal extracted from complex production systems, evaluation sets that tell us what's actually improving, and retrieval infrastructure that keeps our agents grounded. The data team builds and operates the foundation that makes all of it possible.

The Role

You build the data foundation our AI runs on. You design and operate the pipelines that move data from messy production systems into clean, queryable, training- and retrieval-ready form — feeding model development, evaluation, and the retrieval and memory layers behind our agents.

You own data quality and the practical work of extracting training signal from real, imperfect data, and you make sure the people building models and agents can trust what they pull from.

Tech Stack

• Core: Python, Polars, Pandas
• Large-scale processing: Databricks, Spark
• Storage & query: PostgreSQL, MariaDB, SQL
• Cache & retrieval: Valkey, vector stores
• Orchestration: Workflow tooling (e.g., Airflow, Dagster, Prefect)
• Observability: Logfire, Datadog

What We're Looking For

• 4+ years in data engineering, building and operating production pipelines.
• SQL depth — Comfortable against large relational databases, including modeling and query performance.
• Pipeline / ETL-ELT engineering — Batch and, ideally, streaming; with workflow orchestration tooling.
• Large-scale processing — Databricks/Spark or comparable, plus Polars/Pandas for in-process work.
• Data for AI — Building and maintaining training and evaluation datasets, and retrieval/feature infrastructure for agentic systems (embeddings, vector stores), is a strong advantage.
• Reliability mindset — Data quality, validation, and the operational discipline that keeps pipelines trustworthy.
• Communication — Clear documentation and technical writing for a distributed team.