Ravi Anthapu

As a Principal Graph Consultant at Neo4j, I have assisted multiple customers across various domains and multiple use cases. Here are few projects I was involved with.

Patient Claims Analysis

Built optimal graph data model using patient claims data set (LAAD) dataset to perform drug effective analysis and chronic disease analysis at AstraZeneca. This approach reduced analysis time from few weeks to minutes for 3 years worth of data comprising more than billion claims. Customer published an article about effectiveness of this model.

(https://neo4j.com/blog/improving-patient-outcomes-algorithms-graphconnect/)

Build Digital Twins

Built a digital twin graph model at Lockheed Martin to represent the F35 plane BOM's (As designed, As engineered, As delivered, As maintained) in a single graph to analyze and identify cost overruns due to delays in delivery of parts or identify bad parts being reused across planes and thus causing downtime with usage etc.

Built a digital twin graph model for Cadence systems to represents the chip circuit diagram as a graph and overlay the simulations on it, to identify and address issues in the chip. The analysis time can be reduced from months to less than a day.

Modernizing Customer Loyalty Program

As part of modernizing customer loyalty program, Kroger selected Neo4j graph to store customer profiles and attributes, to retrieve the loyalty card details for the household along with other attributes under 100ms, when a loyalty card screened at a store. I was responsible for graph data model and building optimal queries. For 50 million customers data, system is able to respond under 50 ms, way below SLA.

Enterprise MDM

Built an Item Master MDM system using graph model for Kroger. The item attributes can come from 3 different sources, and each source may provide different information.

· Ingest the Item attributes (simple/complex) into graph at attribute level.

· Ability to define rules to curate the Item data.

· When an Item is modified, all applicable rules are applied to Item source data to create curated version per rule.

· Ability to run rules on ad-hoc basis when the new rules are added, or existing ones are modified.

· This Item master is used to provide website description and legends.

Enterprise Network Architecture

For Wellsfargo built Graph model for end-to-end network connectivity from URL entry point to database servers using the configuration details of load balancers and network traffic data captured by stealth watch. The data flow is captured at the port/IP/Host level. Based on this model a Host-to-Host interaction flow is captured.

· Display host connection path from URL to back end for given URL for security and conformance analysis.

· Provide ability for investigators to look for Paths where traffic is not flowing via a firewall.

· For PCI compliance check if in the network connectivity path any server is not in compliance.

· Simplified the process of network architecture validation and impact analysis of deploying or relocating applications.

Cybersecurity

Built graph model for Government agency to represent the network architecture to analyze the attack vectors. Associate the known CVE threats analysis to the servers to identify the network vulnerabilities and their exposure to outside world.

Customer 360

Built Graph model to do customer 360 tasks for large automotive company.

· Connect customers across different verticals (Cars, Energy etc.)

· Provide Dashboards Ex. Store Manager looking for Cars ready to be delivered in the next week

· Provide Data segmentation views for various departments

· Ability to keep the customer interactions and responses from the company to address the issues in a more effective manner

Worked as senior architect for Group Data Dictionary (GDD) - BCBS 239 Standard Implementation using Neo4J Graph DB. GDD is the metadata repository that contains the data required to describe the data elements and entities with UBS systems across organization. The metadata captured with in GDD will describe data entities in terms of their definition, elements and relationships. The flow of data through the organization will be tracked through lineage at the data entity level.

· Designed the Logical Data model for Neo4J, with Oracle as the source data model.

· Built a reconciliation process to ensure that the data in Oracle and Neo4J Graph DB do not have any discrepancies and are in sync.

· Setup and manage the Neo4J cluster in production environment.

· Lineage process developed is used by Data governance for compliance and audit purposes.

Worked as part of IEEE Collabratec™ (https://ieee-collabratec.ieee.org/) product development as consultant. My role involved in developing the Web UI, mobile app development and back end services for storing the content in Elastic search and Neo4J databases.

• Developed backend REST services to store the message posts, files, question answers in elastic search database and retrieve to show them.
• Store the users, their connections and interactions in Neo4J database to track the individual connections and activity.
• Worked on mobile application development using Cordova, Ionic and AngularJS frameworks.
• Developed Ping OAuth based authentication mechanism for the application access for Web and Mobile.
• Created Multi-tenant framework to be able to save data to MongoDB as primary and Elasticsearch for text search.

Cue-me multi-modal platform is the first multi-modal application development platform based on the W3C MMI architecture (https://www.w3.org/TR/2012/REC-mmi-arch-20121025/) to develop multimodal applications on various platforms including mobiles. Worked as the chief architect of the platform and was responsible for design and development of the platform and its various elements. The core Device Manager is developed on J2EE platform.

· Developed these core elements in native languages (Java, Objective-C, C#) for Cue-me platform

o SCXML (State Chart XML) processor in Java, Objective-C and C# to handle the interactions in the application elements.

o InkML, SMIL parsers and handlers to handle the Ink modality content.

o Remote Speech recognition proxy to avail the speech recognition services on low end devices or where local recognition is not possible.

o Integrated embedded speech engines like Loquendo, IBM embedded speech engines to provide on device speech recognition and text to speech capabilities.

o Integrated Pocket Sphinx speech engine on iOS, Android and Windows Phone platforms for local key word spotting and grammar based speech recognition.

· Setup and Manage the environment for Development, Testing and Release artifacts.

o Setup VMWare hypervisor environment (Linux, Windows VM’s) for testing the various products.

o Setup Jenkins to do builds and push them to VM’s for automated testing.

o Configure Jenkins plugins to do J2EE and Android builds on Linux, Windows Phone and Windows Desktop builds on Windows and XCode builds on MAC.

o Setup Jenkins to run custom load testing clients to perform performance testing Device Manager.

o Setup Jenkins to trigger Selenium clients to perform mobile device testing.

o Product release builds to release artifacts to Maven repository using Jenkins.

Responsible for Cue-me product deployment at Wal-Mart. This is the main platform used to develop cross platform applications for enterprise usage across all stores, distribution centers and the other Wal-Mart associates. The product is deployed in various stores (more than 4000) across the world for local access and also in centralized location in Bentonville for scenarios where local store deployment is not possible. Across all the deployments total 800,000 users are currently on the platform using various applications developed by Wal-Mart and Openstream. The applications are deployed for store use, inventory management, warehouse logistics, payment processing and Sam’s club membership. Played key role in creating a deployment package to be able to push the platform to all the locations with ease. The deployment was certified for 40,000 simultaneous users.

1) Global Online Fulfillment

a) Provided guidance for getting most out of the Cue-me platform to develop this application on desktop, Windows Mobile and Android platforms using Javascript and J2EE technologies.

2) Sam’s club Membership

a) Provided guidance and in the development of the application on iPad devices. Built a module to use the Ingenico payment devices for handling the payments on iPad device. This was developed using Java, Objective-C and C++.

3) BigChief Application

a) Designed and lead the team to build the BigChief application for store associates and managers interact using multi modal content. This was developed using JQuery and REST services and deployed on Websphere.

4) Money Transfer Kiosk Application

a) Designed and lead the team to develop Kiosk application to run on large tablets deployed as kiosk terminals in store. This was developed using Ionic, JQuery and AngularJS frameworks and REST services.

5) Payment Processing Component

a) Designed the Component to be flexible to adapt to various vendor devices along with PCI compliance mechanisms embedded into the component. The PCI key management of rotating the keys on the device to protect the card data is handled by the component.

Responsible for Cue-me product deployment at BNY Mellon. Also, lead the team to design and develop the first mobile application deployment on iPad.

· Lead the team for developing the HTML5 Sencha based mobile application Global Liquidity Services (GLS) for representatives to manage accounts.

· The application was completed in 3 months from design to deployment.

· Developed Sencha widgets for high performance.

· Optimized the performance of the application to make sure the orientation changes are smooth like native transformations. Sencha handling of orientation changes was taking more than 10 seconds to complete the UI rendering. This was brought down to under a second.

Responsible for deployment of the Enterprise Virtual Assistant product to Merck enterprise users.

· Integrate Yammer into Smart Assistant

· Built custom document interaction process flows to share documents securely on mobile devices.

· Personal E-mail assistant to integrate into the enterprise mail.

Responsible for fine tuning the performance of the J2EE application to migrate the data from mainframe to Websphere/Oracle based system.

· Analyze the Oracle query performance.

· Compare the mainframe batch job performances to J2EE batch job performance and identify the tuning areas

· Recommend application/query changes to improve the performance

Developed Stock Swap module for exercising the employee options. When the employee wants to exercise the employee stock options, the employee can either pay the cash for the cost or trade existing company stock to cover the cost of the options. Responsible for developing the web interface and the backend DB logic.

Worked as developer in the X-Windows based application to replace the existing green screen application that the customer representatives using to provide phone connections. Developed MVC paradigm for the UI and backend. When a value was updated by some process any UI elements displaying those values received events to update the UI. The UI side abstracted so that the type of the data is made simple and UI can be developed easily

Worked as Informix developer to develop an utility to perform performance testing. Developed a custom scripting language to define the test scripts and the tool will interpret the scripts in a scalable manner using multi-threading.

Worked as device driver developer for MPE/ix operating system. Worked with adding new features and fix the issues for SCSI disk device drivers and network device drivers and kernel programmer.

Worked as a developer for in house product developed for Software quality assurance. This tool was meant to analyze the C code developed by other teams for software quality. The tool parses the C programs and instrument them to track the execution paths etc. Also, the instrumentation captures the data entered during the normal test and can perform regression test with that test data. My role was

• Build a scanner and parser for C language

• Build a symbol table to hold all the elements provided by the parser.

• Analyze the symbol table for unused variables and functions.

• Build a source tree and give it to instrumentaiton module.

• Generate the final source after instrumentation is complete and generate the binary with instrumented code

This tool captured the code flow during runtime to be able to calculate the code coverage metrics, unused variables, code quality metrics easily. It can also capture the input and output and run the regression tests on the changed code and validate the functionality is not affected.