Shabari Vignesh

Led end-to-end AI-powered merchant analytics platform on AWS. Built a centralized data lake consolidating data from 3+ external systems (Shopify, Fiserv, Clover), processing 10M+ transactions and inventory records daily. Orchestrated LLM inference APIs in production to surface reliable merchant insights; managed Kubernetes clusters, Helm, and Argo CD pipelines for scalable AI platform operations. Implemented observability with Prometheus, Grafana, and ELK; deployed event-driven ingestion pipelines (EventBridge → Step Functions → Lambda) to unify data capture across sources. Reduced data fragmentation by 70% by replacing siloed reports with curated, analytics-ready datasets; improved data freshness from hours to near real-time by shifting batch processes to event-driven processing. Automated reconciliations and led data-quality checks to ensure accuracy. Migrated initial ingestion outputs from JSON to Parquet with curated schemas; improved Athena query performance by 4x-6x. Drove a

Developed demo and evaluation datasets derived from Salesforce, Google Calendar, and Highspot to train internal AI models powering demo experiences. Generated synthetic, realistic sales and calendar data aligned to real structures to speed product iteration by ~30%. Automated data ingestion and transformation workflows, reducing manual prep by ~50% for sales, product, and demo teams. Built Looker dashboards visualizing sales engagement, platform usage, and rep activity; ensured safe, privacy-preserving demos by using synthetic data.

Client: Ellevio. Built scalable ETL pipelines in Azure Data Factory with incremental loads, retries, and monitoring to ingest operational and meteorological data reliably. Implemented SQL transformations to standardize time zones, deduplicate device IDs, and produce analytics-ready datasets. Modeled and optimized Azure Synapse/SQL Server data warehouses using star schemas; enabled near real-time grid signals via Kafka and Spark Streaming pipelines. Enforced data quality checks and governance; improved data freshness and reliability for grid monitoring and operational analytics. Enabled containerized, versioned AI/analytics workflows with orchestration and CI/CD integration; ensured scalability across multiple environments and teams. Also supported client BAT MES project data pipeline onboarding and reconciliation processes, including ETL for global MES and finance data, with backfill, deduplication, and data validation logic; reduced manual rework and improved reporting consistency acr