Anish Goswami

HSBC Wholesale IT - Lead Data Scientist
• KYC Document Classification : Classify Image and Text based KYC & regulatory documents. Developed Tornado microservice which runs OCR text extraction service & ML based classification service which handles 15K request per day.
• For Deployment, custom CI/CD pipeline is built using git webhooks & ansible. Service is being monitored via graphana Dashboard which uses InfluxDB and Telegraph.
• Document Information Extraction : Extraction of specific text from regulatory PDF document's. Used OpenCv4 functionalities for Image processing, and TableNet and FasterRCNN architectures for Region of Interest extraction. ROI parsed to OCR for Text Extraction.
• Fraud Patterns Analyzing : Using Graph Analysis, to identify fraud patterns in KYC data.

CUMMINS INC - GLOBAL ANALYTICS CENTER

• Created a forecasting model to predict the market share of the Cummins engines from macroeconomic data of all the engine manufacturers.
• Built text classifier which classifies text based upon the sentiment score for the Warranty claims reviews for the engines, leading saving in 6.8 million dollars. It had an accuracy of 87 percent.
• Developed Trend Shift algorithm to find recent structural breakpoints where significant slope change occurred which is used to determine top 20 fault codes from 400+ fault codes.

• POC - Video Analytics Helmet Detection - Detecting Workers in the plant not wearing helmet.

• Handling the fraud & abuse problems of e-commerce and ticketing companies based in Latin America.

• Used Spark MLlib to predict fraudulent transactions, spark streaming and kafka as orchestration layer and druid for visualization.

• Understanding the fraud nuances of the business and provide end to end Fraud Prevention Solution to improve profitability.

• Analyzing customer data using R & Python to find patterns and best fraud indicators.

• Quantify Fraud metrics based on Entity Relationship Modelling.

• Build & Validate hypothesis around each metric to decide criticality of variables for fraud detection.

• Database migration from RDBMS to Cassandra.

• On-boarding customers with Simility systems and assist them to deploy a fully automated fraud detection system.

• Successfully optimized and deployed Rule Tuning methods across all clients, this provides precise rule threshold value.

• Service Desk Tickets Vertical Classifier - Analytics performed on text data to automate the process of ticket assignment by saving 22% of manual effort with faster ticket resolution.

• Best in Class Accounts - Finding Best in Class Accounts among EY clients, segmentation of accounts based upon Business Potential and Financial Metrics using 2-stage clustering method.

• Opportunities Prediction - Build a machine learning framework was used to implement a solution to analyze the data and predict if an opportunity would be won or lost.

• Used Apriori technique/ Market Basket Analysis, for better services packaging for customers based on their profiles.

• Assisted Supply chain analysts with automating reporting functionality using Power BI reporting, Dashboards & Scorecards (KPI) and MySQL, AWS & Data warehouse data sources.

• Bridged with the onshore team to gather requirements and develop Business Intelligence Solutions and build Predictive Data Models

• Interfaced with multiple teams to address production debugging and support automation issues

• Worked on a project to identify associations between various insurance products offered to the client through Market-Basket analysis/Affinity analysis to determine the common sequence of product purchases (customer journey) and the ideal/average inter-purchase gap between two products

• Independently handled an end to end project on identifying cross-sell opportunities - Performed a logistic regression analysis and identified the propensity of each customer to buy a specific product.

- Provided an interactive tool which shows the effect of variables (identified through analysis) on the purchase of the product to be cross-sold, an account’s propensity to buy the product and gives relevant contacts for the selected set of accounts

Deutsche Bank

• Migrated 700 HR and IT Reports into Oracle Business Intelligence (OBIEE) from SAP Business Objects.
• SAP BI – Wrote SAPBI queries to find out backlog application tickets. Predicting potential SLA miss and ticket resolution time.
• Weekly SLA metrics analysis in Excel 2010 and develop these reports to present it to management.
• Closely work with clients to raise Enhancements and Fixes based upon Application Ticket Analysis.
• Installed, upgraded & configured database solutions through OBIEE tools; created dashboards, reports and ad hoc queries.