Hire Vinay H. - RAG Engineer

* Generative AI Architecture: Architected an end-to-end Generative AI pipeline using LLMs and Vector Databases to deliver RAG-based, domain-specific Q&A for internal knowledge sources. * Computer Vision: Delivered a Computer Vision PoC for person detection and pose estimation using advanced object detection models, achieving ~90% real-time accuracy. * Tech Stack: Focused on production-grade implementation using FastAPI, LangChain/LangGraph, and Vector Databases.

Architected predictive data pipelines for industrial battery systems, moving from 30-minute intervals to 1-minute real-time pings and reducing fault detection latency by ~40%. Supported system-level validation and field deployments, ensuring reliable data acquisition and smooth coordination with cross-functional engineering teams. Note: Role ended due to company-wide organizational restructuring.

* Trained and fine-tuned deep learning models (YOLO, ResNet, NVIDIA TAO, custom keypoint detection networks) for applications in action recognition, license plate detection, and real-time object detection. * Generated synthetic datasets using GANs to handle data scarcity and improve robustness across edge cases. * Processed and managed large-scale datasets (100K+ images/videos) to boost model accuracy and stability in production environments. * Built and optimized end-to-end ML pipelines, including OCR-based traffic violation detection systems. * Developed REST API endpoints for real-time inference, enabling smooth integration with client-facing applications. * Deployed ML models on edge devices (NVIDIA Jetson family) by setting up inference pipelines, optimizing hardware utilization, and preparing devices for on-site customer deployments. * Conducted model training, evaluation, and testing workflows, ensuring reliable performance across multiple production use cases.