Data Engineering-III

Job Description – Lead Data Modeler, DEX Data Platform Modernization Job Description – Lead Data Modeler, DEX Data Platform Modernization Oct 2, 2026 · @ABHISHEK Role overview Kotak Mahindra Bank is hiring a Lead Data Modeler to own the target-state data model for the modernized DEX data platform on AWS. The role sets modeling standards, designs the enterprise layer of generic (conformed) dimensions and facts, and ensures every data mart is fed from that single trusted layer. Attribute Detail Title Lead – Data Modeling (DEX Platform Modernization) Experience 12–14 years in data; 6+ years in hands-on enterprise / dimensional modeling Location Mumbai Reports to Team Platform Head of Data Leads 4–8 data modelers; partners with data engineering, architecture, governance and business teams DEX on AWS, medallion lakehouse (Bronze / Silver / Gold) Key responsibilities Define the DEX target-state data model, assigning a clear modeling purpose to each medallion layer. Decide where Inmon-style integration (3NF / Data Vault) and Kimball dimensional models apply, and document why. Build the enterprise bus matrix linking business processes to conformed dimensions across retail, corporate, transaction banking, risk and finance. Design generic dimensions (Customer, Account, Product, Branch, Employee, Date, Currency, Channel) and generic fact patterns reused by every mart. Page 1 of 3 Job Description – Lead Data Modeler, DEX Data Platform Modernization Lead design of subject-area data marts (Deposits, Loans, Cards, Collections, Regulatory Reporting) built only from the conformed Gold layer. Migrate legacy warehouse models to AWS with traceable lineage and reconciliation. Produce physical designs for AWS (Redshift keys, Iceberg/Parquet partitioning, Glue Catalog) and source-to-target mappings for engineers. Own naming, key and modeling standards; embed data quality, PII classification and masking (RBI, DPDP Act) into the model. Lead, mentor and grow a team of 4–8 modelers; run model reviews and approve schema changes. Data modeling best practices expected Generic (conformed) dimensions Built once in Gold and reused by all marts; no mart creates its own Customer or Product. Surrogate keys, with natural key and source system retained, plus "Unknown" and "Not applicable" members. SCD type chosen per attribute: Type 1 for corrections, Type 2 for history, hybrid where both views are needed. Role-playing dimensions (e.g. transaction, value and posting date), mini-dimensions for volatile attributes, junk dimensions for flags. Generic facts Grain declared first, in one sentence, before measures or dimensions. Fact type matched to the process: transaction, periodic snapshot (daily balances), accumulating snapshot (loan origination to disbursement) or factless. Atomic grain, narrow tables, explicit handling of semi-additive (balances) and non additive (ratios) measures. Late-arriving facts and early-arriving dimensions handled through inferred members. Feeding the data marts Marts read only from conformed Gold dimensions and facts, never from Bronze or sources. Shared dimensions agreed through the bus matrix so cross-mart reports reconcile. Aggregates added for performance only after the atomic model is correct; lineage, reconciliation and an owner for every mart. Page 2 of 3 Job Description – Lead Data Modeler, DEX Data Platform Modernization Required skills Kimball and Inmon (expert): star/snowflake schemas, bus architecture, SCDs, the four-step design process; Corporate Information Factory and normalized EDW; ability to combine them as a hybrid (integrated Silver feeding Kimball stars in Gold). Data Vault 2.0 is a plus. AWS (working knowledge): S3, Iceberg/Parquet, Redshift, Athena, Glue and Glue Data Catalog, EMR/Spark, Lake Formation, DMS, IAM and KMS. Medallion architecture basics: Bronze for raw, immutable data; Silver for cleansed, integrated data with enterprise keys; Gold for conformed dimensions, facts and marts. Understands CDC, incremental loads, partitioning and schema evolution. Tools: expert SQL; reads PySpark; Erwin, ER/Studio or SqlDBM; dbt and Git; exposure to Power BI, Tableau or QuickSight. Qualifications Bachelor's or Master's in Computer Science, IT, Engineering or related field. At least one end-to-end enterprise warehouse or lakehouse delivery with conformed dimensions and multiple marts. At least one legacy-to-cloud warehouse modernization, AWS preferred. Banking or financial services experience across core banking, loans, cards, payments and treasury data. Good to have: BFSI reference models (BIAN, FSLDM), regulatory reporting (Basel III, Ind AS), AWS or CDMP certification. Clear communicator who holds the line on standards while staying pragmatic on delivery. What success looks like Timeframe Expected outcome 90 days 6 months Modeling standards and layer-by-layer approach signed off; bus matrix drafted for priority domains Core conformed dimensions live in Gold; first two marts migrated and reconciled 12 months Most priority marts served from Gold; schema changes through CI/CD; team of 4–8 fully staffed Page 3 of 3