Jithin Raj Reghuvaran

-Built an analytics platform on Microsoft Fabric, unifying 5 ERP sources using Azure Data Factory, Azure OneLake and 200+ dbt models for 100+ users.

- Scaled voice AI to production for IoT devices, serving 1M+ daily requests with optimized transformer models (DistilBERT-ONNX) on Triton. - Implemented dynamic batching and Redis caching of frequent commands handling 1M+. - Implemented RAG using OpenAI, Pinecone, and LlamaIndex for STB platforms.

- Led development of automated feature stores & ML pipelines for real-time inventory forecasting. - Built production ETL pipelines (Spark, Airflow, Hadoop) unifying CRM, ERP & transactional data into a centralized data warehouse. - Developed a Vector Autoregression (VAR) model deployed as a REST API resulting 40% reduction in equipment downtime via automated failure alerts

- Fine-tuned BERT models to improve clinical negation detection accuracy by 36%, and enabling more reliable diagnostic decision support. - Optimized HL7 note ingestion by 70% through parallelized ETL pipelines and efficient NoSQL design , handling 100K+ daily records for real-time analytics. - Developed chatbot to automate 30% of medical coding workflows, reducing manual effort and coding errors.

- Streamlined sales analytics by automating KPI dashboards (Tableau), enabling 60% faster reporting and data-driven decisions for multi-brand distribution. - Drove 15% improvement in customer retention through data-driven churn analysis and tailored engagement strategies for different retailer segments.