Anurag Shukla

Worked on developing the MLOps platform using microservice architecture in Python. Utilized FastAPI to create individual services for data preparation, feature engineering and model services. Implemented an authentication layer using Okta to ensure secure access to the platform. Integrated CosmosDB as metastore for storing workflows, team information, and other metadata. Used Databricks for enhanced data processing and analytics capabilities within the platform. Applied K-Means clustering and DBSCAN algorithms to analyze Partner data. Developed an image classification model focused on accurately classifying themes in Banner. Integrate image classification model with Azure Form Recognizer API for text detection and with Azure Custom Vision for object detection. Integrate several Generative AI use cases like text classification, summarization, sentiment analysis, chatbots, insight actions, and key phrase extraction using the Azure Open AI service. Developed a language translation model using GPT-3.

Worked on data cleaning, data analyzing, and developed visualization charts and forecasting future value. Developed Fault detection model using XGBoost algorithms and achieved accuracy up to 92%. Worked on Machine Learning Pipeline such as data ingestion, data validation, data transformation, model trainer and model evaluation and reduced the maintenance cost by 20%. Created API using Flask framework and integrated with the application. Used Python library like NumPy, Pandas, Matplotlib, Seaborn, Scikit-learn TensorFlow. Used tools like Anaconda, Jupyter Notebook, Pycharm, Postman, Git. Analyzed model prediction accuracy using classification report, confusion matrix and AUC score. Worked on AWS Services such as EC2, ECR, S3, IAM etc. Developed Automation scripts and workflows.