G
GokulAnanth Sasikumar
About
Detail
Udumalaipettai Municipality, Tamil Nadu, India
Résumé
Jobs
verified_user
0% verified
Software EngineerCognizantJan 2022 - Aug 2023 (1 year 8 months)• Supported and maintained 10+ production Apache Airflow workflows orchestrating AWS EMR jobs, troubleshooting DAG failures, Spark job failures, task retries, and scheduling issues to maintain reliable daily ETL processing for downstream reporting and analytics.
• Investigated production ETL incidents using Airflow and Spark application logs, SQL diagnostics, and AWS monitoring, resolving schema, data quality, and source connectivity issues across 5+ enterprise data sources, including S3, SQL Server, Oracle, and SFTP, minimising disruption to downstream data consumers.
• Performed root cause analysis for recurring pipeline failures and implemented permanent fixes, improving operational reliability and reducing repeated manual inte- Software Engineer InternFULL CreativeJun 2021 - Jul 2021 (2 months)• Deployed a Spring Boot application on AWS Elastic Beanstalk, configuring an Application Load Balancer and CPU based Auto Scaling to support reliable application availability under varying workloads.
• Assisted with AWS IAM configuration by creating and updating policies, roles, and user groups, applying least privilege access controls for application and cloud resources.
• Integrated Amazon SES with a live web application to automate transactional and notification emails, improving the reliability of application-generated communications to end users.
Education
verified_user
0% verified
- Msc. Data Science and EngineeringUniversity of DundeeJan 2023 - Jan 2024 (1 year 1 month)Core Modules: Statistical Modelling, Data Visualisation, Machine Learning, Deep Learning, Big Data Analytics.
- B.E. in Computer ScienceKarpagam College of EngineeringJan 2017 - Jan 2021 (4 years 1 month)
Projects (professional or personal)
verified_user
0% verified
- NYC Taxi Demand Forecasting PlatformPersonal ProjectJun 2026• Built an hourly taxi-demand forecasting pipeline using 10M+ trip records and 50GB+ of raw data, with PySpark handling trip aggregation, lag features, rolling windows, and location-based features.
• Used time-based validation to train and compare XGBoost models, with MLflow tracking the experiments, parameters, and model versions across different feature sets.
• Set up Airflow DAGs for daily feature generation, model retraining, and demand prediction. The final model achieved an RMSE of 3.42 pickups/hour, compared with 4.22 from a moving-average baseline. - Credit Card Fraud Detection SystemPersonal ProjectMay 2026• Worked with 8M+ transaction records to build a fraud detection pipeline, creating 20+ temporal and velocity-based features to capture unusual transaction patterns.
• Trained and compared XGBoost models through 30+ MLflow experiments, adjusting scale_pos_weight and the prediction threshold to improve fraud detection while keeping false positives under control.
• Reached an AUC-PR of 0.86, compared with 0.67 from Logistic Regression, and served the final model through FastAPI in Docker with p99 inference latency below 35ms. - Customer Churn Prediction PlatformPersonal ProjectApr 2026• Built a churn prediction pipeline that pulled user activity data from PostgreSQL and produced daily risk scores for customers likely to leave.
• Used SMOTE-NC to handle the class imbalance and compared models using cost-sensitive evaluation, with MLflow used to track experiments and store model artifacts.
• Packaged the prediction service with Docker and FastAPI and connected it to Airflow for scheduled batch scoring, with the pipeline running successfully on 99.2% of scheduled runs.