Senior Backend Engineer

We are one of the leading technology groups in Italy, with operations in the United States. We operate as a Business & Technology Accelerator for companies, financial institutions, and public sector organizations, supporting the evolution of their IT and digital strategies. Through an ecosystem of companies that share values, business strategies, and a common vision, we enable our clients to unlock the full potential of technology, amplifying its positive impact on both the economy and society. This role is dedicated to our US market. What to expect As our Senior Data Engineer, you own the pipelines that feed our products, which turn core banking data from US credit unions and master data from industrial distributors into analytics, lending and marketing decisions. Between the client’s file drop and the number on the dashboard sits your work: ingesting data that arrives irregular, partial and shaped differently from one client to the next; the shared relational model every credit-union product reads from; transformations expressed as Dagster assets with partitions and backfills; and the checks that say the number is right. You are the data engineering function, working closely with the backend engineers whose APIs serve what you produce and with the data scientists and LLM engineers who consume it. The data is regulated (GLBA, FCRA), so lineage, access control and reproducibility are design constraints, not paperwork. Our stack is PostgreSQL, Dagster, Pandas and Polars, but we hire for the understanding underneath the tool and expect you to carry it across. Your ultimate goal is data that arrives correctly, on time and explainably—and you own it on the morning a client file does not arrive. What you will do
• Ingestion & Integration: Build and harden the file and fileset ingestion pipeline—schema detection and validation, late, missing and re-sent files, idempotent loads, and failure reports that the people who own the client relationship can act on— so that onboarding a new client is a configuration, not a project. Files are the transport today; pulling directly from a client API or source database on a schedule is on the roadmap and yours to design.
• Data Modeling: Own and evolve the shared credit-union data model and the per-product marts on top of it—schemas, migrations, constraints, indexes—so that every product reads one truth and the backend can query it fast.
• Orchestration & Transformation: Define pipelines as Dagster assets with partitions, schedules and sensors, backfills and asset checks, picking the right engine for each step: SQL in PostgreSQL for set-based work, Polars or Pandas for what the database does poorly. Any table can be rebuilt from source, any run can be explained, and a failure is visible before a client sees it.
• Data Quality & Observability: Build the checks—freshness, volume, referential integrity, distribution drift—that turn “the numbers look off” into a specific alert with a specific cause, and own the runbook when they fire.
• Enablement & Engineering Discipline: Give data scientists, analysts and LLM engineers stable, documented datasets, help them turn a notebook into an asset when it has to run every day, and set the conventions for how data work is done here— naming, testing, review, CI on GitHub Actions—and write them down so they outlive any one pipeline. Data code is tested, reviewed and deployed the same way as the backend.
• Regulated Data: Treat the regulated nature of the data as a design input—least-privilege access, PII handling, lineage and audit trail—so that when a client’s due diligence asks where a number came from, the answer is in the system, not in someone’s memory. What we expect from you
• Scope You Have Owned: Experience as a Data Engineer, or as a Backend Engineer on data-heavy systems, in which you have designed and run production pipelines end to end—chosen the model, the schedule and the failure behaviour—rather than only building tasks inside someone else’s design, and set the conventions others then worked to. We weigh what you have owned, not the years it took.
• SQL & Relational Depth: Real fluency in SQL and production experience with one serious relational database—PostgreSQL preferred; SQL Server, MySQL or Oracle count. Transactions, indexing, query plans, window functions, schema design, migrations. You can explain why a query is slow and fix it.
• Pipeline Orchestration: Production experience with a modern orchestrator—Dagster preferred; Airflow, Prefect, Kestra or similar fully count—and, more importantly, the concepts underneath: DAGs, idempotency, partitioning, backfills, retries, schedule versus event triggers, and why an asset-oriented view differs from a task-oriented one.
• Python for Data: Solid Python with a dataframe library—Pandas and/or Polars preferred; PySpark, DuckDB or dbt-style SQL transformation count—including the judgment to push work to the database instead.
• Batch Ingestion of Messy Data: You have loaded data from external parties who did not follow the spec—CSV, fixed-width, Excel, Parquet, JSON, from SFTP or object storage, or pulled from a client API or source database—with encodings, schema drift, duplicates and late arrivals, and you built something that survived the second client.
• Data Modeling: You can design a normalized core model and a denormalized mart on top of it, and argue for when each is appropriate.
• Software Engineering Practice: Git, code review, tests for data code, CI/CD, Docker. You treat pipelines as software.
• Ownership & Communication: You run what you build, you say early when a deadline or a data source is not what it seems, and you can explain to a product manager why a number changed.
• Nice to Haves: Dagster in depth (asset checks, partitions, sensors, IO managers); Polars and the Arrow ecosystem (lazy and out-of-core execution, Parquet, DuckDB interop); SQL Server, which the credit-union products run on today while PostgreSQL is the target; GCP (GCS, Cloud SQL, BigQuery, Vertex AI pipelines) or the equivalent on another cloud; Deploying data workloads as containers on Kubernetes with GitHub Actions; Banking, credit-union or other regulated data (GLBA, GDPR), particularly where a regulation shaped a design rather than producing a document; Master data and entity resolution (deduplication, matching, ontologies, MRO or industrial catalogue data); Feeding ML and LLM workloads (feature tables, training datasets, retrieval corpora); Analytics engineering (dbt, semantic layers, BI tools such as Metabase, Superset or Looker)