Analyzed 4 years invoices for visualization and clustering of the customers, and their payment habits, by identifying 4 customer clusters. First summarized the monthly open invoices information using Python and then predicted next 12 month’s overall overdues and commercial losses at different levels of aggregation using ARIMA models with very high degree of accuracy using R.

The model can be retrained and reused using new monthly inputs, with increased accuracy. The output was put into visualization using a Shiny-app module, by showing the predictions onto a Google-map of Europe, with filters, and facility to download the predicted dataset.

a. Designed the overall architecture of the product and the service APIs to the backend ML product.

b. Developed service for suggesting similar sounding words to the user from existing words dictionary.

c. Developed intent prediction and answer prediction engines for the chatbot, using Stanford’s Glove words representation, Google’s word2vec utilities and cosine similarity. The intent predictor helps the business logic layer to decide which API should be called to get information from other systems and get personalized data. For generic data, the business layer calls the answer predictor AI engine, which returns the best possible answer and the suggested questions to be show to the user.

d. Developed various metrics (like number and trend over a period of time, of interest areas of the users, sessions, comparison of two time-durations, user-feedback) for user-interaction analytics dashboard, for the Chatbot administrators. The log-data is read from DynamoDB on AWS.

1. Developed the long-term product roadmap and full requirement of the product features and format of the visualization charts / tables to be displayed, after brain-storming and business analysis exercises with the SME/s. Managed product backlog.
2. Guided developers for POC and MVP of the product using .Net, Amcharts, Infobright (columner DB for fast yet large datasets) and SQL Server (for user and license management) databases.
3. Designed the multi-tenant database model and the overall application and ETL architecture for extracting and uploading the data from the client system into the product.
4. Developed predictive machine learning models for sales, inventory and a recommender system for suggesting user/s next best report s/he might be interested to view, based on the browsing history. Clustered the customers based on RFM (Recency, Frequency, Monetary) for targeted marketing.
5. Developed licensing and user management processes.

1. Developed the requirement documents (Use case model and Business rules document), including the dashboard visualization requirements, by brain storming with the client-managers.

2. Guided the development team in preparing high-level and low level designs, reviewed and arranged customer sign-off on them. Tracked and managed scope, changes and risks. Prepared project reporting in planned meetings with various stakeholders and at the steering committee meetings.

3. Did change management in collaboration with the client’s IT and Business Project Managers.

4. Did key liaison between the client’s team/s and the India delivery team for effectively identifying, planning, monitoring and closing project risks and issues.

5. I guided both the delivery and client teams for setting-up effective and controlled User Acceptance Test (UAT) through preparation of Acceptance Test Plan and cases documents. I was actively involved in UAT (with the users), defect identification, tracking and identifying and also in detailing forthcoming enhancements and global rollout plan of the application.

1. Developed a team of 8 persons with good working knowledge of Core Java, JSP and servlets.

2. Client engagement in doing requirements study and business analysis. Brain-stormed with the client managers on various features, visualization requirements and architecture of the application - prepared the System Requirement Study (SRS) document.

3. Lead the activities for designing the application on Model View Controller (MVC) architecture in Java and Oracle RDBMS - which includes making of Use Cases, User Acceptance Test plan and cases, System Integration Test plan and effort estimation.

4. Managed scope, analyzed specifications and estimated the project and enhancement requests. Communicated with client on DCRs (Design Change Request) and ROMs (Rough Order of Magnitude).

5. Planned work-distribution to the team members and coordinated among them. Tracked and monitored progress of the project using MS-Project and reported to the senior management as and when required.