Full-time Role | Itorizon Pvt Limited / UCBOS | Bangalore

Overview

As a Data Scientist at Itorizon Pvt Limited / UCBOS, I contributed significantly to various aspects of machine learning and data analysis. My role encompassed deploying neural network models, developing data analysis services, implementing cross-validation techniques, and exploring ensemble approaches for forecasting, regression, and classification modules.

Key Achievements and Responsibilities

  • Neural Network Deployment:

    • Deployed classification and regression neural network models in the TensorFlow framework.
    • Utilized batching techniques for efficient model training.
    • Implemented auto-hyperparameter optimization, including epoch and dropout rate adjustments.

  • Data Analysis Service:

    • Developed robust data analysis services with effective visualization techniques.
    • Applied normalization and transformation operations to enhance data quality.
    • Conducted descriptive statistics for insightful data understanding.

  • Cross-Validation Techniques:

    • Introduced cross-validation techniques for regression, classification, and forecasting modules in both TensorFlow and Pyspark frameworks.

  • Ensemble Approach:

    • Implemented ensemble approaches in the Pyspark framework for forecasting, regression, and classification modules.

  • Explainable AI (XAI):

    • Worked on SHAP (SHapley Additive exPlanations) plots in LSTM models for Explainable AI, enhancing machine learning interpretability.

  • Forecasting Module Leadership:

    • Led the forecasting module, demonstrating a high level of accountability and contributing to ML product development in the R&D department.

  • Model Bias and Data Drift:

    • Worked on developing and implementing modules focusing on Model Bias and Data Drift, highlighting a commitment to addressing critical issues in machine learning models.

Learning and Growth

Throughout this role, I gained valuable insights into machine learning product development, particularly in forecasting modules. The experience provided a deep understanding of neural network deployment, data analysis, and the importance of model explainability and fairness.

This role has significantly contributed to my expertise in the dynamic field of data science, showcasing a comprehensive skill set in both traditional machine learning frameworks and cutting-edge technologies.