Role Summary Hands-on Foundry specialist who can design ontology-first data products, engineer high-reliability pipelines, and operationalize them into secure, observable, and reusable building blocks used by multiple applications (Workshop/Slate, AIP/Actions). You’ll own the full lifecycle: from raw sources to governed, versioned, materialized datasets wired into operational apps and AIP agents. Core Responsibilities - Ontology & Data Product Design: Model Object Types, relationships, and semantics; enforce schema evolution strategies; define authoritative datasets with lineage and provenance.
- Pipelines & Materializations: Build Code Workbook transforms (SQL, PySpark/Scala), orchestrate multi-stage DAGs, tune cluster/runtime parameters, and implement incremental + snapshot patterns with backfills and recovery.
- Operationalization: Configure schedules, SLAs/SLOs, alerts/health checks, and data quality tests (constraints, anomaly/volume checks); implement idempotency, checkpointing, and graceful retries.
- Governance & Security: Apply RBAC, object-level permissions, policy tags/PII handling, and least-privilege patterns; integrate with enterprise identity; document data contracts.
- Performance Engineering: Optimize joins/partitions, caching/materialization strategies, file layout (e.g., Parquet/Delta), and shuffle minimization; instrument with runtime metrics and cost controls.
- Dev Productivity & SDLC: Use Git-backed code repos, branching/versioning, code reviews, unit/integration tests for transforms; templatize patterns for reuse across domains.
- Applications & Interfaces: Expose ontology-backed data to Workshop/Slate apps wire Actions and AIP agents to governed datasets; publish clean APIs/feeds for downstream systems.
- Reliability & Incident Response: Own on-call for data products, run RCAs, create runbooks, and drive preventive engineering.
- Documentation & Enablement: Produce playbooks, data product specs, and runbooks; mentor engineers and analysts on Foundry best practices.
Required Qualifications - 7+ years in data engineering/analytics engineering with 4+ years hands-on Palantir Foundry at scale.
- Deep expertise in Foundry Ontology, Code Workbooks, Pipelines, Materializations, Lineage/Provenance, and object permissions.
- Strong SQL and PySpark/Scala in Foundry; comfort with UDFs, window functions, and partitioning/bucketing strategies.
- Proven operational excellence: SLAs/SLOs, alerting, data quality frameworks, backfills, rollbacks, blue/green or canary data releases.
- Fluency with Git, CI/CD for Foundry code repos, test automation for transforms, and environment promotion.
- Hands-on with cloud storage & compute (AWS/Azure/GCP), file formats (Parquet/Delta), and cost/perf tuning.
- Strong grasp of data governance (PII, masking, policy tags) and security models within Foundry.
Nice to Have - Building Workshop/Slate UX tied to ontology objects; authoring Actions and integrating AIP use cases.
- Streaming/event ingestion patterns (e.g., Kafka/Kinesis) materialized into curated datasets.
- Observability stacks (e.g., Datadog/CloudWatch/Prometheus) for pipeline telemetry; FinOps/cost governance.
- Experience establishing platform standards: templates, code style, testing frameworks, domain data product catalogs.
Success Metrics (90–180 Days) - ≥99.5% pipeline success rate, with documented SLOs and active alerting.
- <20% runtime/cost reduction via optimization and materialization strategy.
- Zero P1 data incidents and ≤4h MTTR with playbooks and automated remediation.
- 3+ reusable templates (ingestion, CDC, enrichment) adopted by partner teams.
- Ontology coverage for priority domains with versioned contracts and lineage.
Example Work You’ll Own - Stand up incremental CDC pipelines with watermarking & late-arrivals handling; backfill historical data safely.
- Define business-ready ontology for a domain and wire it to Workshop apps and AIP agents that trigger Actions.
- Implement DQ gates (null/dup checks, distribution drift) that fail fast and auto-open incidents with context.
- Build promotion workflows (dev → staging → prod) with automated tests on transforms and compatibility checks for ontology changes.
|
