Senior Data Engineer, Agentic Systems
Software Engineering, Data Science
Madrid, Spain
Posted on Aug 28, 2026
Fountain sells agentic software—and intends to run on it too. Our data platform is healthy and owned (dbt and Dagster on Kubernetes; ClickHouse Cloud primary analytics store; CDC off Postgres and MongoDB; many third-party sources; Snowflake/Redshift/BigQuery/object-store lakes). Engineers on the team own models and pipelines; this role owns the agentic buildout. You will own two halves:- How our data team works: design agent-authored models and pipeline changes, agent tooling/context, review and CI workflows, evaluations, and guardrails; reach upstream into product where most data problems start- How the rest of the company works: agentic analytics so non-technical teams can get trustworthy answers (semantic context, skills/MCP surfaces, access boundaries, evaluation, and adoption) What you’ll do:- Design/build the agentic development workflow for the data team (PR/review patterns, CI, guardrails, standards)- Push upstream into product/engineering to shape features and verify data architecture pre-merge- Build the evaluation layer (regression suites, correctness checks, lineage, observability)- Build agentic analytics capability for the company (semantic context, MCP, access controls, query/cost guardrails)- Drive adoption across GTM, Finance, Product, Support; teach patterns; close gaps between design and real usage- Stay hands-on in the platform codebase What you should bring:- 5+ years in data/analytics engineering or equivalent depth- Real depth in analytical data modeling and SQL- Hands-on production experience with dbt and an orchestrator (Dagster/Airflow/Prefect/Temporal)- Strong Python and solid software engineering habits (VCS, testing, code review, CI/CD)- Shipped agentic/LLM-powered systems (tool/function calling, context/retrieval, orchestration, evaluation)- Sound judgment about data governance/PII in multi-tenant environments- Ability to lead change, teach, and persuade without authority Nice to have:- Depth in columnar/MPP warehouses (ClickHouse Cloud; Snowflake/BigQuery/Redshift)- Building with Claude (API, Agent SDK, MCP servers, subagent patterns)- AWS/Kubernetes/IaC; semantic layer/BI (Omni, Looker, dbt SL, Cube); LLM observability/eval tooling (Langfuse, Braintrust); streaming/CDC (Debezium, Kafka, Kinesis); led technical practice change; HR tech domain What success looks like:- Week 1 — shipped a small agent-assisted change to production- Week 2 — first agentic workflow running with safe review/CI/guardrails- Week 4 — broader usage, eval layer catching regressions, first internal agentic analytics live- Week 8 — agent-initiated is default for a class of work; adoption/documentation mature; trustworthy capability no longer bounded by team size