Software Developer in Paris, France

This is an enterprise mobile phone application that was distributed to thousands of staff and customers. The aim was to allow the employees of the client company to receive assignments on the road, report on their activities, and keep the customers informed on the state of the jobs they are paying for.

I designed the architecture and led the development team behind this application.

Customers are kept up to date via an SMS sent from the back-end servers via the Twilio API, while the staff receives notifications through their devices via the AWS Simple Notification Service (plugged into Google Cloud Messaging, and AWS SQS for fanning out to the appropriate collaborating staff).

I designed and built an SMS-based platform on .NET. It enables a client to receive, and process text messages on short-codes, and react to them by performing database queries and responding with text messages localized to the specific region of the original sender.

The entire project was hosted in the AWS Cloud and made extensive use of Lambda, S3, SNS, DynamoDB, Pinpoint, RDS (PostgreSQL), and CloudWatch.

I designed and built a communications platform using Node.js that enables a client to reach out to their customers by email, browser push notifications, SMS, or voice calls.

The project was hosted in the AWS Cloud and used Twilio to make voice calls and send SMS.

An exciting aspect of the project is the innovative use of SNS, SQS, and Lambda to achieve massive scalability while keeping costs under control.

I designed and built a system enabling different service owners to provide a formal description of their service (name, list of endpoints and operations, input parameter schemas, output object schemas) using standard description grammars (WSDL, WADL, and more), authentication schemes, and providers.

In return, the system ensures the discovery, interoperation (with the appropriate sequencing of operations and marshaling of input/output parameters) across protocols and service styles; without the need to write a single line of code.

Utilizing a C++ written native layer that intercepts socket-level events, I designed and built a cross-platform component. The component computes statistics on network performance (errors, port utilization, data rates, and more), buffers them, correlates them to method calls within the code of a currently running a .NET or Java application. It then sends the augmented-stack traces to a remote collection server using an optimized binary transfer protocol.

Using TensorFlow, I wrote code that runs a model to examine past offering, contrasts them with customer engagements (purchases, cancellations, and more), and periods to determine the optimal (leading to the most purchases and least cancellations) product offering on a future date. The resulting code produces an accurate result for dates up to two weeks in the future using a few months of training data