Looking for a job

Amit Kumar Swain

@amitkumarswain

Message

Machine Learning Engineer who built an industrial safety detection pipeline with 87.4% mAP@50:95.

India

Message

At Visakhapatnam Steel Plant (RINL), I engineered a real-time object detection and safety compliance pipeline for multi-angle video feeds, achieving 87.4% mAP@50:95 and a 0.89 F1-score. I also built an OpenCV ingestion and preprocessing pipeline for PPE compliance detection.

At Edunet Foundation, I developed a computer vision classification workflow using MobileNetV2 transfer learning across 90 classes, reducing validation loss by 18%. In my projects, I fine-tuned a T5 text summarizer and built a FastAPI service, and developed a Streamlit loan-risk application that achieved 0.91 AUC-ROC.

Experience

Work history, roles, and key accomplishments

Machine Learning & Computer Vision Intern

Visakhapatnam Steel Plant (RINL)

May 2025 - Jun 2025 (1 month)

Engineered a real-time object detection and safety compliance pipeline processing multi-angle video feeds, achieving an 87.4% mAP@50:95 and 0.89 F1-score across high-noise industrial environments. Built an automated ingestion and preprocessing pipeline in OpenCV incorporating morphological filtering and IoU overlap algorithms to optimize PPE compliance detection accuracy.

OpenCV Object Detection Computer Vision data preprocessing Model Evaluation Benchmarking

Artificial Intelligence Intern

Edunet Foundation

Mar 2025 - Apr 2025 (1 month)

Developed an end-to-end computer vision classification workflow using transfer learning (MobileNetV2) across 90 target classes, integrating data augmentation techniques to mitigate class imbalance. Performed extensive exploratory data analysis (EDA), hyperparameter tuning, and error analysis to reduce validation loss by 18%.

Transfer Learning MobileNetV2 Computer Vision Data Augmentation EDA Hyperparameter Tuning Git