Priyanshu Seth

Results-driven data science professional with experience at Mahindra.AI, where I optimized ETL pipelines and enhanced model precision by 15%. Proficient in SQL and data visualization tools like Apache Superset, I excel in transforming complex datasets into actionable insights, while fostering collaboration across teams.

• GCP Cloud Architecture, https://www.credly.com/badges/bb37dad7-b6e2-481e-a47501f317f162eb/public_url
• Implement CI/CD Pipeline on GCP, https://www.credly.com/badges/f92984e3-5594-4bc1-9f71a231dcfda2ec/public_url

Real Time Accident Severity Prediction (3 weeks)

• Tech Stack: AWS (S3, EC2, IAM Roles, VPC, Sagemaker, Cloudwatch, EMR), Python, Spark, Machine Learning, Big Data Hadoop, Apache Kafka.
• Project involves designing a real-time machine learning system that predicts the severity of road accidents using historical data. The system is primarily aimed at enabling emergency responders to allocate resources effectively based on the predicted severity, ultimately aiming to reduce mortality rates in severe accidents.
• The entire system forms a robust ETL (Extract, Transform, Load) pipeline. Data is extracted from the Kafka topic, transformed into a format suitable for the ML model, predictions are made, and the results are loaded back into the S3 bucket.

Vehicle Price Prediction, (1 week)

• Tech Stack: Machine Learning, Python, Numpy, Pandas,Scikit-learn Jupyter Notebook
• Developed a machine learning model to predict vehicle prices using a dataset of 250K+ rows and 5 key features, extracted through data cleaning and feature engineering from 24 original columns. Implemented and evaluated multiple models, including multiple linear regression, decision tree, and random forest regressors. Achieved best accuracy of 82% with multiple linear regression, outperforming other models (random forest: 78%, decision tree: 76%).