Raveesh Sehrawat

I’m a results-driven Data Scientist & AI/ML Engineer focused on designing scalable ETL pipelines and deploying production-grade machine learning systems. I bring deep expertise in RAG pipelines, NLP, transformer models, and real-time data engineering.

At the Centre for Development of Telematics (CDoT), I profiled and refactored FastAPI endpoints, reducing API latency by 40% and data-fetch errors by 35% on a live government telecom system. I also fine-tuned a pyannote speaker diarization model and engineered a 3-second audio segmentation pipeline, achieving 92% speaker identification accuracy in production.

I led a cross-functional team of engineers to design and deploy React dashboards for real-time API performance metrics, improving page-load time by 20% through a modular FastAPI microservice architecture. I’m similarly hands-on across projects—building a multi-class astronomical classifier with 96.5%+ accuracy, training a CNN digit recognizer to 99% accuracy, and developing an imbalanced churn prediction pipeline achieving 90% accuracy.