Aathithya Arasu S

Machine Learning and RAG engineer focused on building production grade multimodal AI systems that run efficiently on real world hardware. I design and deploy offline Retrieval Augmented Generation pipelines, computer vision workflows, and agentic AI architectures end to end—from data ingestion and vector indexing to LLM serving and evaluation. My experience includes architecting multimodal RAG with hybrid retrieval, confidence scoring, hallucination checks, and A/B testing, as well as document tampering detection pipelines using EfficientNetB7, ELA, and Grad CAM on AWS SageMaker. I work primarily in Python with PyTorch, LangChain, Qdrant, FAISS, and Docker, and have shipped internal tools that improved retrieval quality and debuggability of ML systems. As a B.Tech IT student with strong competitive hackathon exposure and industry internship experience, I combine academic fundamentals with hands on system design, prioritizing robustness, observability, and cost aware deployment.