Data Analysis
Uncover hidden insights and patterns within your data to identify trends, anomalies and opportunities for data-driven decision-making.
Hello, I'm
Manpreet Singh
AI & Data Science Professional
Get To Know More
About Manpreet Singh
I’m a final-year BSc student in Artificial Intelligence and Data Science, with hands-on experience in machine learning, deep learning and data analysis. My academic journey has been shaped by projects involving Chronic Kidney Disease prediction, road image segmentation and interactive data dashboards.
What excites me most about AI and Data Science is the constant evolution, real-world challenges and endless learning opportunities. I believe in the power of intelligent systems and data-driven insights to transform industries, solve complex problems and shape the future. I aim to contribute meaningfully to this journey by building solutions that create real impact.
Beyond data, I indulge in my love for sports, prioritize a healthy lifestyle and find joy in the rhythm of good music. These interests provide balance, fuel creativity and keep me energized for data's challenges.
Here Are My
Proficiencies
My proficiency in technical tools like Python & it's core libraries, SQL, Scikit-learn, TensorFlow/Keras, Open-CV, NLTK and spaCy, combined with strong soft skills, enables me to deliver impactful solutions for real world problems.
Data Wrangling
Clean, transform and prepare messy datasets using Pandas and NumPy. Handle missing data, outliers and data normalization to improve model accuracy.
Data Visualization
Present insights through charts, graphs and dashboards using Matplotlib, Seaborn and Power BI. Create intuitive visual narratives to communicate findings effectively.
Machine Learning
Design and implement models for classification, regression and clustering. Use Scikit-learn and real-world datasets to build predictive systems that improve decision-making.
Deep Learning
Work with neural networks using TensorFlow and PyTorch. Apply CNNs for image tasks and explore sequence models like RNNs and LSTMs for time-series and text data.
NLP
Process and analyze human language data using NLTK and spaCy. Build applications like sentiment analysis, text classification and keyword extraction.
Computer Vision
Extract meaningful patterns from images using OpenCV and deep learning models. Work on tasks like object detection, image segmentation and edge detection.
Model Deployment
Deploy machine learning models using Flask or Streamlit to make them accessible via web applications. Integrate front-end elements for user-friendly interaction.
Collaboration
Collaborate seamlessly with your team to integrate data solutions into your business processes. Whether it's short-term projects or ongoing data-driven initiatives, I'm here to support your success.
This is my
Experience
June 2025 - Aug 2025
Artificial Intelligence Intern -> Punjab AI Excellence
As an Intern at Punjab AI Excellence, I worked on developing an advanced road segmentation system using deep learning, with a special focus on sharpening edge detection for autonomous navigation. I built a U-Net based semantic segmentation model from scratch, created a preprocessing pipeline and converted multi-class masks into binary road masks. To tackle blurry boundaries in conventional models, I introduced an edge-weighted Binary Cross-Entropy loss using Canny edge detection, which boosted mean IoU from 87% to over 92% while maintaining ~97–98% accuracy. I validated the model on real-world road images, contributed to a research paper and strengthened my expertise in image segmentation, loss function customization and applied AI research, with future work aimed at real-time deployment.
Browse Through My
Projects
Brain Tumor Detection
Developed a Brain Tumor Detection system using transfer learning with VGG16 to classify MRI images into three tumor types and a non-tumor case. Implemented fine-tuning, dropout regularization and data augmentation for robust performance. Achieved high accuracy and AUC through detailed evaluation using ROC curves, confusion matrices and classification reports. Built using TensorFlow, Keras, OpenCV and Scikit-learn, demonstrating expertise in deep learning, computer vision and medical image analysis.
Road Segmentation
As part of my AI & Computer Vision learning journey and growing under the mentorship of Dr. Sandeep Singh Sandha, I developed a road segmentation system using the U-Net architecture on the CamVid dataset, focusing on enhancing edge detection for safer autonomous navigation. Building a segmentation model from scratch helped me deeply understand U-Net architecture and I gained hands-on experience with image preprocessing, mask generation, and custom loss functions. I plan to further optimize this system for real-time applications and explore advanced models like DeepLabv3+ or SegFormer.
Human Attribute Detection
Developed an advanced AI-driven visual analysis system using Google Gemini (models/gemini-2.5-flash) integrated within a Streamlit web interface. The model performs multimodal inference to extract detailed human attributes — including gender, age estimation, ethnicity classification, and emotional state — directly from uploaded images. Implemented a custom prompt-engineered pipeline for structured attribute generation and optimized real-time performance through efficient API handling and dynamic image preprocessing. This project demonstrates expertise in AI integration, API utilization and full-stack ML application development.
Credit Card Fraud Detection
Developed a machine learning-based fraud detection system to identify suspicious credit card transactions from highly imbalanced data. The project utilized the Credit Card Fraud Dataset, containing anonymized PCA-transformed features (V1–V28) alongside transaction time and amount. Applied data preprocessing, exploratory data analysis (EDA), and class imbalance handling to enhance detection accuracy. Implemented and evaluated multiple models including Logistic Regression, Random Forest, and Gradient Boosting, measuring performance using metrics like precision, recall, F1-score, and ROC-AUC. This project demonstrates skills in data analytics, feature engineering, model optimization, and fraud risk prediction.
Chronic Kidney Disease Prediction
This project predicts Chronic Kidney Disease (CKD) using a machine learning model trained on clinical data from 400 patients. It uses features like blood pressure, serum creatinine, hemoglobin, and more to classify patients as CKD or not. A Random Forest Classifier was trained by comparing with several other models and achieved strong accuracy, F1 scores and cross-validation. The model is deployed through a Flask web application with a clean user interface. This tool aims to support early disease detection and assist healthcare professionals in diagnosis.
Pizza Sales Analysis
Designed and developed an interactive Power BI dashboard to analyze pizza sales data from January to December 2015. The dashboard provides deep insights into key performance metrics such as total revenue (817.86K), total orders, average order value, and pizzas sold. It identifies sales trends by day, month, category, and size—highlighting that Classic pizzas and Large sizes generate the highest revenue. The report also showcases best- and worst-performing pizzas by revenue, quantity, and total orders, enabling data-driven decision-making for inventory and marketing strategies.
Check out my
Publications
Explore My
