Saman Qazi

Summary

Aspiring Data Scientist with experience in machine learning, NLP, data analysis, and Al-driven solutions. Skilled in Python, SQL, and analytics, with practical experience in working with complex datasets. Seeking to apply analytical skills to drive data-driven decision-making

Overview

year of professional experience

Certification

Work History

Data Sciene Intern

IIT Jammu | Kaleidofin Private Limited

06.2023 - 01.2024

Credit risk analysis on the real world company's credit data -KALEIDOFIN PRIVATE LIMITED
Explored data trends using various techniques, including K-means clustering, PCA, Autoencoder, Bruesh-Pagan, Yeo-Johnson transformation
Developed and optimized predictive models, starting with Logistic Regression, and later implementing Support Vector Machines (SVM) and hybrid ensemble models (SVM + Perceptron), significantly enhancing model performance.
Successfully improved model performance by leveraging advanced feature selection techniques (MRMR, WoE, Toad, SHAP) and deploying hybrid models, leading to a measurable increase in predictive accuracy, AUC score, and recall rate. These improvements directly enhanced the company’s ability to assess credit risk more accurately, reducing the likelihood of defaults and optimizing risk management strategies.

Education

B.Tech - Computer Science

Jammu University

J&K

08.2023

High School Qualification -

Mallinson Girls School

Skills

Data Science and Analytics
Python
SQL
Machine learning

Azure
TensorFlow
NLP

Websites

Certification

AZ-900: Microsoft Azure Fundamentals, Microsoft
DP-900: Microsoft Azure Data Fundamentals, Microsoft
Databases and SQL for Data Science with Python , IBM / Coursera

Accomplishments

1- Customer Churn Prediction System – [Python, Pandas, Scikit-learn, XGBoost, SHAP, Streamlit]

• Built an ML pipeline using Telco data to predict customer churn based on service usage and account features
• Achieved 0.89 ROC-AUC using a tuned XGBoost model and extracted churn drivers with SHAP explainability
• Deployed a Streamlit tool to score churn risk and support customer retention strategies in real time

2- AI-Powered Resume Parser & Job Matcher – [Python, Spacy, Hugging Face, TF-IDF, Streamlit]

• Built an NLP tool to extract structured details from resumes using NER and pattern matching

• Applied TF-IDF and BERT embeddings to rank resumes by job description similarity (cosine-based)

• Deployed a Streamlit interface for real-time matching; achieved ~85% accuracy and 40% relevance gain over keyword search

3- Personalized Movie Recommendation System – [Python, Scikit-learn, Surprise, Streamlit]

• Developed a hybrid movie recommender using content-based filtering (TF-IDF) and collaborative filtering (SVD)
• Processed movie metadata and user ratings to generate personalized suggestions with improved relevance
• Deployed a Streamlit app for users to input preferences and get top movie recommendations with posters

Timeline