Likith Kumar Dundigalla

● Led migration of 800 tables from SQL Server to PostgreSQL, averting $5M in potential licensing costs. ● Upgraded Databricks and moved to managed storage, cutting costs by 11% and speeding up pipelines by 20% ● Refactored pipelines and moved configurations from DynamoDB to S3, reducing infra costs by 31%.

● Built and maintained ETL pipelines; delivered dashboards for KPI tracking and operational decisions.

● Engineered an AI companion to support mental health, with video responses achieving 75% satisfaction. ● Built a multimodal AI pipeline using Google speech-to-text and Gemini, prompt-engineered for responses under 20secs. ● Implemented Chain-of-Thought(COT) prompting, reducing illogical responses by 70% compared to base models.

● Automated content tagging for 1000+ learning materials using Gemini, reducing manual work by 40%. ● Accelerated concept retrieval by 30% via Neo4j graph with BERT embeddings, cutting student search time. ● Built RAG chatbot for topic exploration, increasing learning engagement by 25% among students.

● Built a full-stack web app to automate university catering event planning and resource management. ● Developed data pipelines to sync event details with staff scheduling and kitchen inventory. ● Implemented a RAG on PostgreSQL, allowing non-technical users to query data using natural language.

● Modernized batch-processing to real-time pipeline cutting processing time by 95% under HIPAA. ● Developed automated checks and Datadog monitoring, reducing outages by 40% ensuring data integrity. ● Wrote SQL Stored Procedures to automate data transactions, ensuring accurate data for downstream teams. ● Scaled the end API to handle 1000+ EHR requests/minute with 99.9% uptime. ● Designed a Kafka/RabbitMQ proof-of-concept to enhance messaging system flexibility and scalability.

● Engineered Spark workflows to unify 50+ event data sources, powering executive-level analytics dashboards. ● Migrated data from DocumentDB to PostgreSQL, reducing storage costs by 25% and querying ease. ● Built interactive DOMO dashboards with embedded data quality checks, eliminating 99% of reporting delays. ● Implemented metadata documentation in DataHub, improving data lineage tracking and easy user onboarding.