Nagaraju Javisetty

• Built and deployed a production-grade agentic AI system using RAG and LLMs, delivering secure, low-latency responses for finance-related customer queries. • Built an OCR + ML classification pipeline to automatically extract structured data from insurance and finace documents—eliminating 100% of manual document review for operational teams. • Developed and deployed FastAPI-based microservices, containerized via Docker, and orchestrated production deployment on Kubernetes with CI/CD workflows.

• Built an AI system to detect industrial leakages by combining infrared and RGB imaging, improving early fault detection accuracy. • Managed the entire process, involving data collection with bi-spectral cameras and applying techniques such as camera calibration, and image registration to align IR and RGB data. • Successfully fused RGB and IR sensor data, trained the image processing model with 4-channel image data, deployed the model. • Tools: Python, PyTorch, OpenCV, Hikvision & Testo cameras

• Labelling vehicle data from endurance runs and measurement campaigns. Creating unit tests for KPI scripts. • Analysing and visualising vehicle data. Scripting for comparing resimulation results with measurement recordings • Developed a neural network for pedestrian and cyclist detection achieving 85% accuracy. • Tools: Python, TensorFlow, Keras, Git.

• Implemented a YOLO-based real-time object detection system on Nvidia Jetson platforms for vehicle and pedestrian tracking. • Integrated Grafana dashboards to monitor traffic detection data and performance metrics. • Tools: OpenCV, IMX477 Camera, Git.