Pankaj Kumar

Passionate data scientist skilled in advanced techniques such as machine learning, statistical modeling, NLP, and computer vision. Excel at transforming raw data into actionable insights that optimize business processes. Goal is to develop innovative, data-driven solutions and contribute to advancements in AI.

Enhancing Safety and Hazard Management in Scrap-Based Liquid Steel Production

Client: India's leading Manufacturer and supplier of TMT Bars.

Business Problem:

Conduct comprehensive risk assessments to identify potential hazards in the scrap- based liquid steel production process.

Business Solution:

• Collected and annotated image data of hazardous materials, including cylinders, canisters, shock absorbers, fire extinguishers, and motors, using Roboflow's Smart Polygon tool and manual polygon annotation.
• Applied various augmentation techniques to the training set to enhance model robustness and accuracy.
• Trained a high-performance object detection model using Google Colab with GPU acceleration, achieving the highest accuracy among tested models.
• Deployed the trained model using Streamlit for real-time hazard detection and visualization.

Technology Stack:

Environment: Google colab 

Annotation Tools: Roboflow

Deployment Tools: Spyder , Streamlit

Business Benefits:

• Enhanced worker safety by 40% by accurately identifying and eliminating hazardous materials before they reached the blast furnace.
• Increased furnace lifespan by up to 25% by preventing damage from explosive materials, leading to significant reductions in maintenance costs and downtime.

• Professional Program on Data Science and AI: 360DigiTMG, 2024
• Python Programming Language: 360DigiTMG, 2024
• Data Visualization using PowerBI: 360DigiTMG, 2024
• SQL Course: 360DigiTMG, 2024