ML Engineer

Design and build Snowflake Semantic Views that define logical tables, relationships, dimensions, facts, and metrics for core network domains and datasets. Translate raw network counters into standardized KPIs aligned with 3GPP measurement definitions Model time-series behavior correctly. This includes granularity (5-min, 15-min, hourly, daily), time-zone handling, late-arriving data, and aggregation rules for additive, semi-additive, and ratio-based KPIs. For ratio KPIs, this means re-aggregating the numerator and denominator rather than averaging ratios. Build conformed dimensions such as network element, node type, vendor, region/site, time, and service/slice, so KPIs can be sliced consistently across domains. Author and maintain synonyms, descriptions, and sample questions so the semantic views work well with Cortex Analyst and natural-language querying. Work with the data engineering team to understand the ingestion architecture (e.g., Snowflake Kafka Connector, Snowpipe Streaming, Openflow, Dynamic Tables, Streams, Tasks etc.). Ensure the semantic layer is built on stable, well-defined interfaces rather than volatile raw tables. Implement automated drift detection covering schema changes (new, renamed, or dropped columns, and data type changes), new network elements or counters, and changes to KPI formulas. Use monitoring, Snowflake Alerts, and event-driven checks. Define and enforce data contracts between ingestion and semantic layers, including versioning, change notification, and backward-compatibility rules. Monitor data freshness and latency SLAs so that semantic views reflect near-real-time data. Configure Dynamic Table target lag or equivalent mechanisms appropriately. Manage semantic view definitions as code, with CI/CD pipelines for deployment across dev, test, and prod. Implement automated testing for KPI logic, including unit tests on formulas, row-count and null checks, threshold/anomaly tests, and regression tests against golden datasets. Maintain a KPI catalog and business glossary documenting each metric’s definition, formula, source counters, owner, and change history. Optimize query performance and cost through clustering, warehouse sizing, materialization choices, and monitoring of compute consumption. Work closely with data engineers on ingestion design and with BI/analytics teams (Power BI, Tableau, etc.) who consume the semantic layer. Act as the point of contact for questions about metric definitions, resolving discrepancies across teams. Requirements

Semantic Views and/or Cortex Analyst semantic models, Dynamic Tables, Streams and Tasks, Snowpipe/Snowpipe Streaming, and performance tuning. Proven experience building semantic layers or metrics layers. Relevant tools include Snowflake Semantic Views, dbt Semantic Layer/MetricFlow, AtScale, LookML, or Power BI semantic models. Solid dimensional modeling skills (Kimball star/snowflake schemas) and experience modeling high-volume time-series data. Working knowledge of telecom core network architecture familiarity with performance management counters and KPIs, is strongly recommended. Understanding of streaming data concepts: Kafka topics and partitions, event time vs. processing time, late and out-of-order data, and exactly-once semantics. Exposure to Python/Snowpark for automation and data quality frameworks. Strong analytical and communication skills, with the ability to explain technical metric logic to non-technical stakeholders.