Priyanka Marihal

Led development of TalentMatch RAG Assistant for LLM-powered candidate-to-job matching; designed end-to-end embedding and indexing pipeline, storing embeddings in Databricks Vector Search for semantic retrieval; implemented similarity search and prompt orchestration to generate explainable recommendations (skills, gaps, match score) and accelerate recruiter decisions. Also contributed to CNH Industrial/UCR Analytics by building predictive models on vehicle telemetry to reduce downtime, optimized Spark jobs with Delta Lake caching and partition pruning to cut cloud costs by ~25% and runtime by ~30%, and supported data governance migrations to Unity Catalog. Worked on Metaverse Chatbot as Senior Developer, implementing NLP-based chat flows and integration with GCP services.

Defined hotspot detection approach for Michelin’s large-scale data; read and processed Parquet data to extract features; performed preprocessing and quality checks; trained Isolation Forest and clustering models for anomaly/hotspot detection; validated model performance and prepared it for production deployment.