HARIKRUPA VEDERE

Built and managed bronze-silver-gold lakehouse platforms using Delta Lake and Microsoft Fabric, enforcing ACID guarantees, schema governance, and zone-based data quality across 15M daily patient records.

Designed and optimized multi-source ETL and ELT pipelines handling batch, incremental, and CDC-based data movement with watermarking, schema drift handling, and dependency scheduling at scale.

Integrated EHR systems using FHIR and HL7 standards, building ingestion pipelines with Apache Kafka and Apache NiFi to consolidate patient records across hospital endpoints and reduce physician data retrieval time by 40%.

Implemented HIPAA-compliant data security including patient de-identification, column-level encryption, and role-based access control across clinical data platforms.

Improved analytical query performance by over 45% through Delta Lake optimization techniques including Z-ordering, partition pruning, and automated table maintenance orchestrated via Airflow.

Built GenAI-powered data assistants using LLMs and RAG architectures, reducing report generation time from days to hours for clinical and business stakeholders.

Deployed Agentic AI workflows using LangGraph and Bedrock Agents to automate complex multi-step clinical data enrichment tasks, cutting manual curation effort significantly per workflow cycle.

Delivered end-to-end retail analytics consolidating Salesforce, Shopify, and SAP data using PySpark across cloud environments, driving 20% better inventory distribution and 30% improvement in demand forecasting accuracy.

Built clinical and operational BI dashboards in Power BI with row-level security, connecting gold-layer lakehouse tables to support high-concurrency reporting across regulated cloud environments.

Designed Azure Data Factory pipelines for CDRs and network logs into Azure Data Lake Storage, applying watermarking and dependency scheduling for data freshness.

Built PySpark jobs on Azure Databricks to aggregate CDR data into call success rates and data consumption measures, using broadcast joins for strict SLA adherence.

Implemented event-driven ingestion with Apache Kafka, Azure Event Hubs, and Google Pub Sub for high-frequency network signaling events, tuning partitioning and consumer groups for throughput.

Utilized Apache HBase for near real-time network metric storage and integrated Apache Druid for fast OLAP queries across 500 billion telecom events monthly.

Optimized BigQuery analytical datasets through partitioning, clustering, and scheduled refreshes, reducing dashboard load times via materialized views for multi-terabyte workloads.

Orchestrated GCP data workflows using Cloud Composer with Airflow DAGs for Pub Sub ingestion, Dataflow streaming, and BigQuery queries, implementing retry logic and SLA sensors.

Implemented CI/CD pipelines using GitHub Actions and Terraform to automate deployment of Azure/GCP infrastructure and Spark jobs, applying Pulumi for infrastructure definition.

Built centralized monitoring using Azure Monitor, Google Cloud Monitoring, and Grafana to track pipeline latency and Spark job performance, integrating Prometheus for tuning.

Built real-time transaction ingestion pipelines using Apache Kafka and Confluent Kafka, configuring partitioning for fault-tolerant delivery under peak transaction volumes.

Designed Apache Flink streaming jobs with Flink SQL and CEP patterns to detect complex fraud scenarios, enabling sub-second suspicious transaction flagging.

Implemented dynamic risk scoring combining rule-based fraud signals with ML model outputs managed by MLflow, achieving 94% accuracy in identifying suspicious transactions.

Built batch ingestion pipelines using Google Cloud Dataflow and Apache Beam to move 5+ years of mainframe transaction records into Google Cloud Storage, sustaining over 10,000 messages per second.

Optimized BigQuery datasets through partitioning and clustering, reducing fraud trend query runtimes from hours to minutes for risk reporting.

Developed Spring Boot RESTful fraud alert APIs with OAuth 2.0 authentication, improving API response times by 60% for external banking systems.

Implemented GCP IAM access controls with fine-grained dataset-level permissions across BigQuery, Cloud Storage, and Cloud SQL for financial regulatory compliance.

Reduced manual fraud investigation workload by over 40% by integrating Kafka, Flink, Redis, HBase, MLflow, and Tableau into a unified fraud operations platform.

Built centralized payroll data migration pipelines using Talend ETL and Azure Data Factory, supporting full historical and incremental loads for over 650,000 client businesses.

Processed and standardized payroll data with Python and Azure Databricks (PySpark), applying Delta Lake merge operations to reconcile migrated calculations.

Re-architected ETL orchestration by migrating dependency logic to Apache Airflow DAGs on Azure, reducing failed payroll runs by 78% during peak cycles.

Optimized Azure SQL Database for curated payroll datasets, improving average query execution time from 3.2 seconds to under 0.4 seconds.

Implemented SOX and PCI-compliant security controls using column-level encryption and row-level security across ADLS, Azure SQL Database, and Azure Databricks with Azure Key Vault.

Developed ML models on Azure Databricks (PySpark ML, Scikit-Learn) to improve payroll cash flow forecasting, achieving 89% precision 30 days ahead.

Applied Apache Spark batch processing in Azure Databricks to migrate quarterly tax filing workloads with ACID-compliant Delta Lake operations, reducing end-to-end tax processing time from 18 hours to 4.5 hours.

Executed phased migration strategies, verifying 100% record accuracy against source systems with 0 disruptions during critical tax filing deadlines.

Built ETL pipelines using Oracle Data Integrator, Informatica PowerCenter, and Talend Open Studio, integrating data from Oracle, MySQL, MongoDB, and Amazon S3.

Set up Oracle Database storage architecture alongside Amazon RDS, Amazon S3, and Redshift, applying partitioning and S3 tiering to improve query performance.

Used Informatica PowerCenter to extract data from MySQL, Oracle, & MongoDB, staging output into Amazon S3 while preserving referential integrity across 12 ERP systems.

Automated ETL job scheduling using Oozie workflow definitions & Cron jobs in Oracle Enterprise Manager, ensuring correct execution order for interdependent processes.

Configured Amazon CloudWatch alerts for Talend jobs processing CSV/XML files from on-premise systems and Amazon S3, enabling rapid incident response.

Used Informatica to cleanse, enrich, and validate enterprise asset management datasets before loading into Amazon Redshift, integrating AWS Glue crawlers for schema discovery.

Built Oracle Business Intelligence reports from Oracle data warehouses and Amazon Redshift data marts using star schemas, designing role-based access controls.

Monitored pipeline execution through CloudWatch for real-time manufacturing performance and supply chain KPI analysis, maintaining audit trails.