EXL—Databricks Tech Lead

Job description

Description: Databricks Platform Engineering ● Design, build, and maintain Databricks workspaces, clusters, and compute pools across dev/test/prod environments. ● Configure and manage Databricks Unity Catalog for data governance, access control, fine-grained permissions, and data lineage. ● Optimize cluster configurations — instance types, auto-scaling policies, spot/preemptible nodes — for cost and performance. ● Implement workspace-level best practices: folder structures, access controls, secret management (Databricks Secrets / Azure Key Vault / AWS Secrets Manager). ● Manage Databricks jobs, workflows, and multi-task job orchestration with dependency management. Delta Lake & Lakehouse Architecture ● Design and implement Delta Lake tables with appropriate partitioning, Z-ordering, and file compaction (OPTIMIZE / VACUUM). ● Build Medallion Architecture (Bronze / Silver / Gold) layers for structured data lake organization. ● Implement Delta Live Tables (DLT) pipelines for declarative, reliable ETL/ELT with built-in data quality expectations. ● Manage schema evolution, table versioning, time travel, and Change Data Feed (CDF) for incremental processing. ● Design data lakehouse patterns integrating Delta Lake with external systems (Kafka, ADLS, S3, GCS). EXL Service 10 Exchange Place, Suite 2200, Jersey City, NJ T: +1.201.748.4700 www.exlservice.com Data Pipeline Development (PySpark / SQL) ● Develop scalable batch and streaming data pipelines using PySpark, Spark SQL, and Delta Lake. ● Build structured streaming pipelines for real-time ingestion from Kafka, Event Hubs, and Kinesis into Delta tables. ● Write optimized PySpark transformations leveraging broadcast joins, adaptive query execution (AQE), and dynamic partition pruning. ● Create reusable transformation libraries, utility frameworks, and pipeline templates for team productivity. ● Implement robust error handling, retry logic, and dead-letter queue patterns in production pipelines. MLflow & AI/ML Workloads ● Set up and manage MLflow tracking servers, experiment registries, and model lifecycle management on Databricks. ● Support data scientists and ML engineers in deploying model training and inference workloads on Databricks clusters and GPU instances. ● Build feature engineering pipelines using Databricks Feature Store for reusable, versioned ML features. ● Enable GenAI workloads — LLM fine-tuning, RAG pipeline development, and vector search (Databricks Vector Search / Mosaic AI). ● Implement MLOps practices: model versioning, A/B testing, model serving via Databricks Model Serving endpoints. Cloud Integration & DevOps ● Integrate Databricks with cloud-native services: Azure Data Lake Storage (ADLS). ● Build and maintain CI/CD pipelines for Databricks notebooks and jobs using Azure DevOps, GitHub Actions, or GitLab CI. ● Implement Databricks Asset Bundles (DABs) or Terraform for infrastructure-as-code (IaC) deployment of Databricks resources. ● Manage data ingestion using Auto Loader, COPY INTO, and partner integrations (Fivetran, dbt, Airbyte). ● Monitor pipeline health, cluster utilization, and costs using Databricks system tables and cloud cost management tools. Governance, Security & Optimization ● Implement row-level security, column masking, and dynamic data views using Unity Catalog policies. ● Ensure data quality enforcement using Delta Live Tables expectations and Great Expectations integrations. ● Conduct performance tuning — query plan analysis, caching strategies, Photon engine enablement. ● Maintain data cataloging, metadata management, and data lineage tracking within Unity Catalog. ● Document architecture decisions, runbooks, and operational guides for Databricks workloads. Requirements: Databricks Platform Engineering ● Design, build, and maintain Databricks workspaces, clusters, and compute pools across dev/test/prod environments. ● Configure and manage Databricks Unity Catalog for data governance, access control, fine-grained permissions, and data lineage. ● Optimize cluster configurations — instance types, auto-scaling policies, spot/preemptible nodes — for cost and performance. ● Implement workspace-level best practices: folder structures, access controls, secret management (Databricks Secrets / Azure Key Vault / AWS Secrets Manager). ● Manage Databricks jobs, workflows, and multi-task job orchestration with dependency management. Delta Lake & Lakehouse Architecture ● Design and implement Delta Lake tables with appropriate partitioning, Z-ordering, and file compaction (OPTIMIZE / VACUUM). ● Build Medallion Architecture (Bronze / Silver / Gold) layers for structured data lake organization. ● Implement Delta Live Tables (DLT) pipelines for declarative, reliable ETL/ELT with built-in data quality expectations. ● Manage schema evolution, table versioning, time travel, and Change Data Feed (CDF) for incremental processing. ● Design data lakehouse patterns integrating Delta Lake with external systems (Kafka, ADLS, S3, GCS). EXL Service 10 Exchange Place, Suite 2200, Jersey City, NJ T: +1.201.748.4700 www.exlservice.com Data Pipeline Development (PySpark / SQL) ● Develop scalable batch and streaming data pipelines using PySpark, Spark SQL, and Delta Lake. ● Build structured streaming pipelines for real-time ingestion from Kafka, Event Hubs, and Kinesis into Delta tables. ● Write optimized PySpark transformations leveraging broadcast joins, adaptive query execution (AQE), and dynamic partition pruning. ● Create reusable transformation libraries, utility frameworks, and pipeline templates for team productivity. ● Implement robust error handling, retry logic, and dead-letter queue patterns in production pipelines. MLflow & AI/ML Workloads ● Set up and manage MLflow tracking servers, experiment registries, and model lifecycle management on Databricks. ● Support data scientists and ML engineers in deploying model training and inference workloads on Databricks clusters and GPU instances. ● Build feature engineering pipelines using Databricks Feature Store for reusable, versioned ML features. ● Enable GenAI workloads — LLM fine-tuning, RAG pipeline development, and vector search (Databricks Vector Search / Mosaic AI). ● Implement MLOps practices: model versioning, A/B testing, model serving via Databricks Model Serving endpoints. Cloud Integration & DevOps ● Integrate Databricks with cloud-native services: Azure Data Lake Storage (ADLS). ● Build and maintain CI/CD pipelines for Databricks notebooks and jobs using Azure DevOps, GitHub Actions, or GitLab CI. ● Implement Databricks Asset Bundles (DABs) or Terraform for infrastructure-as-code (IaC) deployment of Databricks resources. ● Manage data ingestion using Auto Loader, COPY INTO, and partner integrations (Fivetran, dbt, Airbyte). ● Monitor pipeline health, cluster utilization, and costs using Databricks system tables and cloud cost management tools. Governance, Security & Optimization ● Implement row-level security, column masking, and dynamic data views using Unity Catalog policies. ● Ensure data quality enforcement using Delta Live Tables expectations and Great Expectations integrations. ● Conduct performance tuning — query plan analysis, caching strategies, Photon engine enablement. ● Maintain data cataloging, metadata management, and data lineage tracking within Unity Catalog. ● Document architecture decisions, runbooks, and operational guides for Databricks workloads. Responsibilities: Databricks Platform Engineering ● Design, build, and maintain Databricks workspaces, clusters, and compute pools across dev/test/prod environments. ● Configure and manage Databricks Unity Catalog for data governance, access control, fine-grained permissions, and data lineage. ● Optimize cluster configurations — instance types, auto-scaling policies, spot/preemptible nodes — for cost and performance. ● Implement workspace-level best practices: folder structures, access controls, secret management (Databricks Secrets / Azure Key Vault / AWS Secrets Manager). ● Manage Databricks jobs, workflows, and multi-task job orchestration with dependency management. Delta Lake & Lakehouse Architecture ● Design and implement Delta Lake tables with appropriate partitioning, Z-ordering, and file compaction (OPTIMIZE / VACUUM). ● Build Medallion Architecture (Bronze / Silver / Gold) layers for structured data lake organization. ● Implement Delta Live Tables (DLT) pipelines for declarative, reliable ETL/ELT with built-in data quality expectations. ● Manage schema evolution, table versioning, time travel, and Change Data Feed (CDF) for incremental processing. ● Design data lakehouse patterns integrating Delta Lake with external systems (Kafka, ADLS, S3, GCS). EXL Service 10 Exchange Place, Suite 2200, Jersey City, NJ T: +1.201.748.4700 www.exlservice.com Data Pipeline Development (PySpark / SQL) ● Develop scalable batch and streaming data pipelines using PySpark, Spark SQL, and Delta Lake. ● Build structured streaming pipelines for real-time ingestion from Kafka, Event Hubs, and Kinesis into Delta tables. ● Write optimized PySpark transformations leveraging broadcast joins, adaptive query execution (AQE), and dynamic partition pruning. ● Create reusable transformation libraries, utility frameworks, and pipeline templates for team productivity. ● Implement robust error handling, retry logic, and dead-letter queue patterns in production pipelines. MLflow & AI/ML Workloads ● Set up and manage MLflow tracking servers, experiment registries, and model lifecycle management on Databricks. ● Support data scientists and ML engineers in deploying model training and inference workloads on Databricks clusters and GPU instances. ● Build feature engineering pipelines using Databricks Feature Store for reusable, versioned ML features. ● Enable GenAI workloads — LLM fine-tuning, RAG pipeline development, and vector search (Databricks Vector Search / Mosaic AI). ● Implement MLOps practices: model versioning, A/B testing, model serving via Databricks Model Serving endpoints. Cloud Integration & DevOps ● Integrate Databricks with cloud-native services: Azure Data Lake Storage (ADLS). ● Build and maintain CI/CD pipelines for Databricks notebooks and jobs using Azure DevOps, GitHub Actions, or GitLab CI. ● Implement Databricks Asset Bundles (DABs) or Terraform for infrastructure-as-code (IaC) deployment of Databricks resources. ● Manage data ingestion using Auto Loader, COPY INTO, and partner integrations (Fivetran, dbt, Airbyte). ● Monitor pipeline health, cluster utilization, and costs using Databricks system tables and cloud cost management tools. Governance, Security & Optimization ● Implement row-level security, column masking, and dynamic data views using Unity Catalog policies. ● Ensure data quality enforcement using Delta Live Tables expectations and Great Expectations integrations. ● Conduct performance tuning — query plan analysis, caching strategies, Photon engine enablement. ● Maintain data cataloging, metadata management, and data lineage tracking within Unity Catalog. ● Document architecture decisions, runbooks, and operational guides for Databricks workloads.