Rajat Sharma is an expert data scientist specializing in classification, clustering, and anomaly detection. He possesses a strong background in developing predictive models for network security systems, utilizing skills in Python, Pandas, and machine learning. His expertise includes feature engineering, hyperparameter tuning, ETL pipeline development, and data pre-processing.
Experience
Data Scientist
Whizhack Technologies
Developed a LightGBM model to identify and prevent 27 types of security threats and achieved an F1-score of 79%. Performed feature engineering and hyperparameter tuning on 20 million plus records. Impact: Model was integrated into the security system, reducing false positives by 21%. Developed a Random Forest Classifier to identify and prevent 11 types of security threats in network traffic events, achieving 79% accuracy trained on 7 million plus events, using Python and AWS S3. Impact: Replaced a rule-based system, increasing real-time inference speed by 60%. Built K-Mode and DBSCAN clustering models, dividing linked devices with accuracy scores based on the silhouette coefficient (0.71 and 0.68). Impact: Lateral threat prevention by 61%. Built Python ETL pipeline with AWS S3, MongoDB, and Pandas for JSON data, improving scalability and decision-making with millions of daily records. Impact: Real-time analysis from seven sources.
Data Analyst
Dew Solutions
Operationalized client's data inventory, mapped data flows, and ensured compliance with data privacy laws (GDPR, LGPD, CCPA). Conducted DPIA readiness assessments and organized company data for enhanced efficiency and security. Generated analysis reports for stakeholders using Pandas and SQL, providing valuable insights. Leveraged Pandas and NumPy to analyze sales data, identifying patterns, trends, and peak sales months.
Data Science Intern
SoftoBiz Technologies
Developed an efficient data processing workflow using Pandas, SQL Server, and MongoDB. Leveraged Pandas for data manipulation and analysis, SQL Server for structured data storage, and MongoDB for handling unstructured data. Enabled seamless integration and streamlined data operations for improved efficiency and decision-making. Impact: Automated dynamic pricing and intuitive dashboards enhanced pricing strategy and decision-making, driving revenue optimization and business performance for Carlist.my and carsales.au.
Education
Dr. APJ Abdul Kalam University
Bachelor of Technology
Software Development, Python Programming, Artificial intelligence, Database Management