ROSHAN ROY GEOFFREY JOE

AI-driven engineering student with hands-on experience building machine learning models and LLM-powered automation tools. Strengths include the ability to prototype, analyze data, and deliver meaningful solutions. Passionate about the applications of AI, with a focus on experimenting and solving real-world problems across software, data, and robotics.

• AI Programming with Python, Udacity, 11/2025, Neural networks, classifier design, error optimization, argparse ML pipelines
• Azure AI Engineer Associate, Microsoft, 10/2024, Designed and deployed enterprise-grade LLM and computer vision systems

End-to-End Computer Vision Classifier (PyTorch)

• Built an image classifier using transfer learning, which had high accuracy on a custom dataset.
• Implemented full pipeline, including data preprocessing, training loops, evaluation, and validation.

Custom LLM Knowledge Assistant (Azure + Python)

• Built a retrieval-augmented chatbot using Microsoft Copilot Studio and Azure tools for company-specific information.
• Integrated idea flows and prompt engineering for improved response accuracy.

Robotic Arm Control System (Python + PSoC)

• Programmed servos and UART communication to move a robotic arm based on the position retrieved from camera vision.
• Programmed real-time integration between software outputs and connected robotic hardware.

Remote Heart Rate Estimation System (Computer Vision + Signal Processing)

• Developed a real-time, contactless heart rate estimation system using webcam-based face detection, chrominance signal processing, and frequency-domain analysis.
• Applied band-pass filtering and FFT peak detection to extract pulse signals and estimate BPM under normal indoor lighting.