Devarchana Dev Choudhury

Background

About Devarchana Dev Choudhury

Data Science enthusiast with over 4 years of experience in predictive analysis, dataset analysis, and offering solutions for servers in the Tech industry. I focus on transforming raw data into actionable insights for informed decision-making using machine learning algorithms and data mining: • Programming languages: R, Python (Pandas, Seaborn, Sklearn, NumPy, Matplotlib, Pytorch, ggplot2, PySpark), SQL, MATLAB • Deep Learning: Large Language Models (LLM), Hugging Face Transformers, TensorFlow, Keras • Machine Learning: Stochastic optimization, Data Exploration and Mining, Recommender systems, Apriori, Matrix Factorization, MLOps, Linear Regression, Logistic Regression, Decision Tree, Cross Validation, Support Vector Machines (SVM), Gradient Boosted Machines, Feature engineering • Natural Language Processing (NLP)- Text Processing, Generative AI; • Forecasting - Time Series Analysis, ETS/Exponential Smoothing • Methodologies: Agile • Database: Oracle Apex, MySQL, MongoDB • Framework: Apache Spark • Visualization: Tableau, Power BI, Excel, Microsoft Visio • Amazon Web Services: SageMaker, S3, Redshift, Glue, Lambda, EC2, RDS, Bedrock • Microsoft Azure: Synapse Analytics, Data Lake Storage, SQL Database • Others: Web scraping, Git, Salesforce, Hypothesis Testing, Segmentation Analysis, Customer churn, Large and complex data processing : • Designed a predictive maintenance system for HPE GreenLake servers using a Random Forest model with hyperparameter tuning (grid search), achieving 0.91 AUC-ROC, preventing 80+ server failures/year, and reducing outage-related costs by 20%, letting me integrate machine learning solutions with tangible business impact. • Optimized server resource planning by integrating ARIMA time-series forecasting with Bayesian Neural Networks (TensorFlow) for demand uncertainty estimation. Reduced resource provisioning errors by 15%, ensuring consistent server availability: I am a solution-driven professional with a proven track record of identifying critical challenges and delivering innovative, impactful results through strategic problem-solving and cross-functional collaboration. ’ : If you would like to exchange ideas or learn more about my background, let’s connect on LinkedIn. You can also reach me at devarcha@buffalo.edu.