100_CSE_Priya Magesh

I’m an AI/ML Engineer focused on building practical systems across NLP, computer vision, and RAG—always with an eye on measurable impact. Right now, I’m shaping my B.E. in Computer Science & Engineering (Honours in AI/ML) while shipping prototypes that translate quickly into real workflows.

At Omega Healthcare, I built a proof-of-concept AI voice bot with Rasa + Twilio to automate candidate engagement and flag joining-risk signals, tackling a 30%+ drop-off issue flagged by HR. I also integrated GCP Speech-to-Text and Natural Language APIs to drive real-time intent detection, plus an NLP sentiment pipeline that outputs joining-probability scores from interaction signals.

In computer vision, I delivered a content-based jewelry recommendation system using OpenCLIP embeddings with Top-K cosine similarity retrieval under 1 GB VRAM. I benchmarked multiple ViT architectures, tested retrieval robustness across viewpoint changes, and proposed a hybrid metadata + embedding retrieval approach.

I’m equally motivated by end-to-end automation: at Blue Marble Smartware, I built a PDF-to-Excel pipeline using PaddleOCR for Tamil multilingual tables and orchestrated the full loop with n8n (PDF → OCR → structured Excel) with no manual steps. My projects—like Fabric-AI’s RAG-guided synthetic road hazard generation with BlenderProc + FAISS and a YOLOv8 sim-to-real validation—reflect how I engineer for performance, not just demos.