Hire Harshitha O. - Fine-Tuning Engineer

* Architected ML feature stores in collaboration with data scientists, standardizing feature extraction pipelines across 5M+ patient * records and accelerating model training cycles by 50%. * Established LLM fine-tuning data preparation workflows using PySpark on Azure Databricks, processing 10TB+ daily healthcare * claims and reducing data preprocessing time for AI models by 40%. * Designed RAG-compatible ingestion pipelines using REST APIs and microservices, automating 500K+ records/hour into Azure * Data Lake and enabling retrieval-augmented generation for clinical decision support tools. * Optimized Delta Lake infrastructure using ACID transactions, partitioning, and Z-Ordering, improving ML model query * performance by 3x and cutting data latency from hours to minutes. * Deployed CI/CD pipelines for production AI/ML applications, resolving 95% of incidents within SLA and increasing model release * velocity by 25%.

* Engineered Spark-Hive ML data pipelines on Hadoop processing 2TB+ daily banking datasets, reducing batch runtime by 35% * and delivering clean, model-ready features for enterprise AI applications. * Integrated SQL-based data validation and quality frameworks across 10+ ML pipelines, reducing training data errors by 60% and * preventing $500K in downstream model rework for banking AI clients. * Led zero-downtime migration of legacy on-premise systems to Azure Data Lake, processing 1M+ daily transactions and * establishing scalable cloud infrastructure for ML and generative AI workloads. * Applied performance tuning techniques including caching and broadcast joins, achieving 98% pipeline uptime and reducing ML * pipeline incident resolution time by 50%.