Chuma Nnolim

I initiated this project as a means towards a much larger goal of a functional motion-based model. This diagnostic tool provides the capabilities to search and navigate from one part of the body to another. It uses SQL common table expressions (CTE) to navigate through a hierarchical pair recursively.

The hierarchical pairs consist of 3D points extracted from vertices of 3D object files representing parts of the body and matched up as pairs where vertices meet. Each body part exists as part of multiple hierarchies within the Foundational Model of Anatomy (FMA).

There are options for avoiding different body part categories in the search. For example, I can avoid any paths that involve going through the skin when going from the tongue to the occipital lobe.

The search is asynchronous and happens in parallel in three ways:
- A task that traverses hierarchical pairs from the start node towards the end node.
- A task that traverses hierarchical parts from the end node towards the start node.
- A task that monitors the first two tasks can also end the search if the search goes beyond two minutes.

Using the cancellation tokens concept, any task can quit the search if any of the tasks meet their goal.

Re-tooled and re-platformed the existing VB6 system and database to an intranet ASP.NET application. The Public Safety divisions in Minnesota use the system to order and receive items from the Public Safety Warehouse.

• Implemented a preordering functionality to improve self-service adoption for the public safety division admins, especially trooper districts, and removed the dependency on the data entry staff who manually entered orders received via fax from divisions around the state.
• Integrated the system with Speedee shipping system.
• Incorporated internal active directory groups as application groups within the app, controlled by warehouse admins for application data and object security.
• Incorporated an "act as" functionality, allowing warehouse admins to ensure application groups and their users only see data and pages they were entitled to see.
• Incorporated expiration of inventory items sent to all public safety personnel with notifications on upcoming expirations.
Implemented a workflow to allow warehouse staff to control and manage how orders came into inventory, including triggering backorders as needed.
• Added the notifications seen only by admins to review the items requiring immediate attention.

Designed and implemented the replacement of the OpenText Process360 document management application. Vendor stack needed to be decommissioned while still allowing LOB to retain its robust functionality and integration with PeopleSoft procure-to-pay solutions.

-Re-architected the vendor-delivered document capture to use existing enterprise Kofax Capture push interfaces for fax, email, and scan.
-Vendor-delivered document storage was re-architected to use an intermediate Filenet web service interface to the enterprise Filenet storage.
-Replaced the vendor-delivered document workflow with a new custom C# workflow engine, using Windows Services, XML, and SQL tables.
-77 servers were tech-refreshed with significantly reduced VMs, reducing server footprint and costs while increasing application redundancy.
-Migrated FINRA documents from vendor repository to Filenet to meet FINRA WORM compliance for the company with both systems online at the same time without impacting document workflow.
-Rebuilt and redeployed a custom PeopleSoft browser for the accounts payable line of business from VB6 to C#, using Chrome APIs.
-Added PeopleSoft Financials application as a document capture source for the enterprise document centralization.

Implemented an OBIEE analytics solution for the supply chain department to better manage its suppliers and contracts and spend from requisition-to-payment distributions.

The federation included all spending related-data sources such as Emptoris contracts, PeopleSoft procure-to-pay solutions, and spreadsheets used to analyze expenses and invoice images.

Various hierarchies such as product, supplier, people, general ledger, and accounting were included to provide one view of all expense-related data. Communications during invoice processing were included to identify process improvements and improve SLAs.

The default model came with PeopleSoft EPM (SCM). My involvement for each requirement agreed on with the supply chain line of business:
• Modeled existing warehouse schema in OBIEE to enable creating user dashboards.
• Modeled new schema objects from sources with the ETL team for the best representation in OBIEE.
• Ensured correctness and speed of daily dashboard queries.
• Recommended optimizations to the ETL team and end-user report writers and reseeded cache in OBIEE.

I provided PaaS services for other LOBs within corporate finance departments needing reporting or analytics with my team.

This project was for a mining company (Toptal client) that ingests telemetry data for mining devices via a cloud REST API called IXON. The data was needed for reporting in multiple systems.

Since my client was on the business side of the house with limited IT support for their initiatives and had no server infrastructure, I developed a stand-alone Node.js executable to perform data downloads on-demand via the IXON APIs. The application could eventually be ported to a server for scheduling should the bandwidth for IT support increase. The data was stored in a SQL Lite DB database to maintain data integrity and also cumulatively exported for every run. The SQLite database was useful for creating a single data source for all devices and maintaining the integrity of daily downloads.

Eventually, once the target BI platform was revealed, I created a version of the Node.js application as an Iron python data source so that it could be initiated by the platform with instructions to do so should the client decide to take that route.

In addition, I was able to contact the IXON team on behalf of the client and resolve some discrepancies between their documentation and their REST API interface.

This project was for an Automatic Vehicle Driving company (Toptal client) that improves the accuracy of its automatic vehicle (AV) by identifying and labeling objects in the real world captured by its AV. Labeling activities are performed by external vendors worldwide and generate quality auto scores per project, which are absorbed within the company and presented back to the vendors to help improve the quality of labeler activities. Personnel spent 20 hours a week running Looker reports and sending scores to vendors via spreadsheets. The data was stored in a transactional database, reporting tool was google data studio, and using a google connector had poor performance.

I transformed all the auto scores and Looker reports from the OLTP DB into a data source in BigQuery. Then, I developed one Google Data Studio dashboard for all vendors using vendor emails to provide row-level security. Each score has quality expectations that help show how accurately the labelers or projects perform. A Google app was created using Google Apps Script to allow the configuration of expectations for each score in Google Sheets. A BigQuery JavaScript function was used to transform each score to color-code the auto scores for the vendor dashboard.