Saketh G Email & Phone Number

Milwaukee, Wi, Us

Milwaukee, Wi, Us

• Developed a time-based utilization model that calculates the status of the surface mining machine (Wheel Loader) whether it is in productive, non-productive, idle, shutdown modes. • Supervised development of an analytic which calculates similarity score (Dynamic Time Warping) between signal/metrics, helped in classifying/improving cycle states/times, and identifies machine abuse activity (Wheel Loader).• Responsible for design and implementation of scalable python-based analytics (parallel notebook execution using multi-threading concept) in Azure Databricks.• Published PowerBI datasets with custom measures and daily/weekly/monthly Snowflake views, doing this reduced processing time to load visuals/graphs in PowerBI reports by 40%.• Adopted Pytest framework to write unit/integration tests that mock databases which were called by regular Python code blocks.• Collaborated with stakeholders in different regions and supported them with resolutions, onboarding to analytics teams, part of “make use of data” campaigns etc.

Milwaukee, Wi, Us

• Developed an end-to-end machine learning project scoped for streaming (Confluent Kafka & Docker Services), batch framework context.• Created an auto-training unsupervised model (Databricks MLFlow) that selects the best-trained models (Logistic Regression, SVM, Gradient Boosting) over time and applies for them future data.• Autotuned hyperparameters for ML models with the help of MLFlow experiment tracking and model management services, resulting in an accuracy of > 95%.• Clustering algorithms were used to detect outliers/anomalies which in turn helped to label the dataset.• SMOTE oversampling methods were adopted for imbalanced datasets, that happened to cure overfitting and precision/accuracy problems.• Built python-based data analytics that runs through time-series data, identifies productivity of mining machines (worth more than ~100million$) over the lifetime, and produced those results in the form of PowerBI reports for product development/reliability engineering team.• As part of exploratory data analysis, thorough data cleaning, feature extraction, normalization techniques, and pre-processing methods were applied to datasets.• Involved in key design elements for ML job deployment in Azure Databricks (Azure CI/CD pipeline and PowerShell scripting).• Initiated a PySpark grouped map job (Azure Databricks) to run data analysis on historical datasets comprise of ~billion rows, multi-metric, and entire mining fleet IoT sensor data. Followed by storing outcomes in Azure DeltaLake and Azure BLOB storages.• Built python wheels out of analytics and deployed as a scheduled job in Databricks environment and managed using Databricks CLI.• Contributor for internal build python package index (koala) contains abstractions and interfaces to local databases and REST API’s.• Frequent views/tables (Snowflake, Hive, Impala, KUDU) using SQL were created based on the aggregates generated by analytics which supports downstream PowerBI datasets and reports in all.

Milwaukee, Wi, Us

• Fulfilled all data science duties as an intern for a high-end mining firm. • Created and responsible for building machine learning models for accurately categorizing anomalies of sensors in mining machines. • Improved the performance of the current models by 25% through precisely predicting the faults caused by sensors. • Developed visualizations for datasets (>million data points) in statistical graphs and information graphs which helped co-workers to get an idea about the status quo. • Semi supervised learning algorithms are introduced to the work platform for handling inconsistent data.

Kansas City, Mo, Us

Teaneck, New Jersey, Us

• Trained in ASP.NET MVC web application framework with MS SQL procedures for web development. • Responsible for testing client i.e., Schneider Electric's products end-to-end using automation tools by following AGILE methods. • Built, tested, and deployed scalable, highly available, and modular software products. • Performed regression and compatibility testing for products to work on all supported operating systems and software that co-exist with the product. • Automated scripts using .NET framework to run test environment. • Optimized service procedures to increase customer satisfaction.

Master Of Science (Ms), Computer Science

Master Of Science (Ms), Computer Software Engineering