Meet Joshi

Strategic Consultant and Analyst Goal-driven Analyst polished in managing and breaking down large volumes of information. Proactive at heading off issues in operations, workflow and production by uncovering trends affecting business success. Over 3 years of experience in IT industry. Detail oriented analyst with 3 years of experience conducting analysis and research to come up with solutions to common business problems. Polished in formulating business improvement strategies and overseeing new technology and system implementation. Commercially-aware professional with in depth knowledge of computing systems and project management techniques.

Realized Accuracy Calculator (RAC)

• Create script for deleting old data from forecasting-metrics table using DynamoDB with AWSlambda CDK.
• For deleting the rows scan whole table and find rows which do not have TTL and update TTL torows. DDB will delete them automatically when time reaches TTL. This time reaches every 30 days. Additionally, Developed the command line for this script.

Private Pricing 

• Implemented multithreading logic in the back testing API script.
• Recorded the execution time for a fixed number of requests with the thread value of 2, 5, 10 and how it was initially with the single thread.
• Implemented the long-term fix of calling the back testing script via AWS services instead of dev-desktop. Additionally, developed the command line for this logic.

Credit Card Fraud Detection 

• Directed a team of four to assess dataset of 284,807 transaction utilizing python SVM to find fraud on credit card.
• Developed system for that ad attained 94.3% accuracy detected 492 frauds from the dataset.
• Delivered product solution to institutional and corporate clients, done balanced model performance tuning.

Telco Customer Churn 

• Predicted which telecom service (phone service, internet service, streaming TV, Online Backup) used frequently with an accuracy of 82%
• Utilized Python for running xGBoost, Random Forest, SVM, and ADA Boost to figure out the churn.
• Reduced the cost of the project to 20% by using smart devices set up and making the system fault-tolerant and scalable.

Predict request on Sulekha 

• Led a team of 4 to predict and analyze from which area most requests are made to use the product on the website.
• Applied Machine Learning algorithms SVM, Random Forest, and ANN in RStudio and achieved an accuracy of about 86% for prediction.
• Utilized Google Analytics, AdWords and merged them with the prediction model to get the companies with the highest rating and their contact information.