Vishvajit Mohite

Led end-to-end development and production deployment of a Generative AI solution using Azure OpenAI with Retrieval-Augmented Generation (RAG) for automotive complaint classification (Client: Stellantis), achieving ~95% accuracy across 27 regulatory categories and reducing manual review effort by ~40%. Built and productionized ML forecasting models (Random Forest, XGBoost) to predict recall completion rates and parts demand, achieving R² = 0.82 with MAPE < 18%, enabling improved inventory planning and data-driven budget allocation. Designed schema-aware retrieval pipelines and prompt orchestration workflows to ensure reliable LLM-based reasoning and structured database query execution in enterprise production systems. Architected and maintained a highly available AWS RDS platform with encryption at rest/in transit, optimized indexing, and automated backups, reducing query latency by 30% under production load. Engineered scalable AWS Lambda-based data workflows integrated with RDS, S3, and Secrets Manager, achieving 99.9% SLA uptime for real-time accident and recall data processing. Developed Spark and PySpark-based ETL pipelines for large-scale data transformation and feature engine

Built automated data ingestion and web scraping pipelines using Python (BeautifulSoup, Scrapy), increasing data availability by 70%. Developed supervised ML models using scikit-learn, reducing manual analysis effort by 60%. Automated data preprocessing, cleaning, and validation workflows to support analytics and ML use cases.

Conducted exploratory data analysis (EDA) using FADS and GADS frameworks to identify trends and operational bottlenecks. Delivered KPI reports and dashboards that improved operational efficiency and management decision-making by 20%.