Rahul Bhasker

Water Disaggregation

• Developed a tool to identify and isolate different type of water usage patterns utilizing the hourly consumption data.
• Incorporated Domain expertise by adding the functionality of classifying the isolated pattern by the user. This information adds an additional layer of training data for the model.
• Helped utilities to enhance their water capacity planning.

Irrigation meter predictor

• Developed a tool to isolate and identify irrigation events in water consumption. Utilized time series data manipulation techniques along with clustering to recognize clustering events.
• Extrapolated the results of the tool to classify meters as irrigation heavy meters.
• Achieved a detection accuracy of 97% during the QA phase.

Anomaly Detection using Matrix Profiles

• Developed an anomaly detection algorithm using the matrix profile technique to identify potential issues in smart meter pressure sensor readings (time series data).
• Correlated detected anomalies with geolocation data to pinpoint areas with a higher likelihood of pipe system problems enabling proactive maintenance and cost savings.
• Achieved a detection accuracy of 94% during the QA phase.

Sales Prediction with Machine Learning

• Leveraged historical purchase data and customer-specific information to predict sales for key accounts over the next year.
• Implemented a Naive Bayes algorithm using machine learning libraries to generate sales forecasts achieving an 87% accuracy rate for predicted sales for the first half of the year, improving sales team efficiency and resource allocation.

LLM-powered Ticket Summarization Tool (Gen AI)

• Developed a tool to analyze grievance system tickets and identify the top 10 recurring issues over a specified duration.
• Employed Retrieval-Augmented Generation (RAG), large language models (LLM), and prompt engineering to achieve accurate results.

User-Friendly Clustering Exploration Dashboard

• Created an interactive dashboard for clustering analysis, enabling end users to perform clustering without in-depth technical knowledge through simple dropdown selections and clicks.
• Facilitated deeper investigation to isolate root causes of meter-specific issues.

Demographic Consumption Data Analysis Dashboard

• Created a comprehensive dashboard to analyze customer consumption data in relation to demographic factors like average house prices and income.
• Integrated consumption data with US Census and IRS data sets for insightful analysis.
• Managed data cleaning for approximately 17-18 million data points, ensuring data integrity and accurate results.

Iperl Meter Performance Monitoring Dashboard

• Developed a Statistical Process Control (SPC) chart-based dashboard for real-time monitoring of Iperl meter errors,analyzing large volumes of data.
• Implemented CI/CD (Continuous Integration/Continuous Delivery) to streamline UI and back-end deployments,reducing website update time for UI changes from 3 hours to 20 minutes, enhancing operational efficiency.

Data Science

• Created classification model using given historical data of transactions of consumers for prediction.
• Applied Light Gradient Boosting classification algorithm with an with an f1 score of 0.96.

Automation

• Created Design document and got it approved from the Design authority of Client.
• Lead a team of Uipath Developers while playing the role of an Automation analyst and scrum master to create End-to-End Flow of the program, solved all the logical errors and delivered the solution within pre-defined and tight timelines.

Image processing

• Worked with KofaxCapture (Image Processing Tool) to read and extract data from structured Documents by using Advance Zone Locator.
• Enhanced utility (C#) by creating window service to transform output of KOFAX to XML file.
• Enhanced utility (C# ) to migrate documents(8 million) from Legacy System(File Net) to Online repository(Alfresco).