KARTHIK PALYAKERE SURESH

Data Scientist Intern/Master Thesis at Merck Group KGaA, Darmstadt

Currently working on Master's thesis titled "Automation of Segmentation of Nanoparticles in Electron Microscopic Images", focusing on scalable and accurate deep learning-based segmentation workflows. Built multi-step pipelines using Python, PyTorch, and cloud resources (AWS EC2/S3). Delivered reproducible experiments and results tracking for stakeholders. Created high-quality ground truth datasets for microscopic materials including SiO₂ and TiO₂ using high-resolution electron microscopy images.

Research Assistant at Deutsches Forschungszentrum für künstliche Intelligenz GmbH (DFKI), Kaiserslautern

Orchestrated multi-sensor integration systems, improving reliability by 40%. Implemented real-time object detection using PyTorch and ROS2, achieving 90% accuracy. Developed and optimized AI models through advanced fine-tuning techniques. Created efficient data processing pipelines for real-time sensor data analysis. Integrated sensor fusion algorithms for comprehensive object detection.

Data Scientist Intern / Master Thesis at Merck Group KGaA

Developed computer vision pipelines for SEM/TEM image analysis (segmentation, preprocessing, augmentation) improving detection accuracy by 20%. Generated synthetic datasets with advanced augmentation, reducing model overfitting and increasing generalization on limited datasets. Fine-tuned instance segmentation models (Mask R-CNN, YOLOv11, CO-DETR, SAM2), achieving state-of-the-art results for microscopy applications. Automated ML workflows (training, evaluation, deployment) using PyTorch, Docker, and Python, cutting research iteration cycles by 30%. Applied foundation models for feature extraction & domain adaptation, enabling cross-domain generalization.