Gian Piero Bandieramonte

November First is a payment platform for transferring funds internationally in different currencies. The platform has been in place in the company's IT base since I began this project. The goal was twofold: to upgrade the platform to exploit a more comprehensive array of functionalities from the newer version of the FIX protocol and to revamp the platform as it presented deep performance and scalability issues.

I recreated the payment platform in a way that it could perform quote and trade requests, as well as process the respective responses in a faster and more stable fashion. The code was also rendered more maintainable and scalable so that further enhancements could be done more efficiently. A wider array of functionalities was added to the trading platform, including handling requests for quotes (RFQs) and market data simultaneously, enabling the trading of SPOT, TOD, and TOM orders, and supporting express payments.

The offering of these new functionalities solicited by customers and the fueling of the number of transactions they perform on the platform led to a significant increase in new account openings and, consequently, in the company's revenue by at least 12%.

Worked for Brainlab, a company that had an algorithm that performed image registration between X-rays and digitally generated X-rays (DRR). Such an algorithm lacked a unit testing framework to detect potential errors in its precision. Providing this framework was complex because the algorithm relied on 3D imaging APIs such as DirectX. This fact implied that any conformance test verifying the correctness of the algorithm would have to rely on methods that do not use 3D imaging APIs. I therefore designed an image registration method that emulates some parts of the DirectX functionality. Here, geometry was exploited by also using supporting C++ headers. This code performed image registration on specific border cases. The resulting images were then compared pixel by pixel against the results of the original algorithm.

With this method, we found significant discrepancies in the pixel patterns between the algorithm's results and those from my unit testing method. Handling such discrepancies increased the algorithm's precision by roughly 8%, leading to better beam precision during radiotherapy targeting tumors in patients with cancer and to better cancer treatment solutions offered by Brainlab.

I'm currently developing an eCommerce web app for a global organization to facilitate the online sale of books in various formats, including physical and digital delivery.

The app comprises three projects:
• A React storefront that powers the front end with the interface where users can browse and place orders
• A React dashboard where managers can oversee the inventory, fulfill orders, manage users, and more
• A Python API that stores and provides inventory and user data and powers the required back-end functionality

Both the storefront and the dashboard communicate with the API to receive and store inventory and user data. This project is based on Saleor's successful open source platform (Saleor.io/).

Project details under NDA.

I created a web app as a full-stack developer that enabled the online trading of stocks within a brokerage organization that could otherwise only accept trading orders from its clients by phone. This was critical for boosting the organization's rate of orders placed by its clients, ultimately increasing its revenues considerably.

The project was developed using ASP.NET and C# for the back end, SQL to manage the data, and a combination of CSS, HTML, and JavaScript for the front end. Both MS Visual Studio and Adobe Dreamweaver were utilized as IDEs.

Co-founded an investment management firm where I served as its chief analyst and developed an algorithm to cherry-pick undervalued fixed-income securities.

To accomplish this, financial data from hundreds of companies were downloaded through the Bloomberg platform using its proprietary API, which was fed into the Excel models I created. An algorithm ran using such data powered by Excel VBA to aid in analyzing huge amounts of data. An array of charts was also created to assist in the process of selecting the undervalued financial securities.

In addition to picking undervalued fixed-income securities, they also had to be tailored to the financial needs and goals of our clients for whom we designed the investment portfolios. The aforementioned algorithm, at the same time, was responsible for this sort of tailoring.

The project's goal was to implement a program that finds the optimal path to solve any shuffled instance of Rubik's cube using an acceptable amount of time and memory resources. By optimal path, I mean the shortest possible path from the initial shuffled instance to the ordered instance. There can be multiple paths leading to the same ordered instance of Rubik's cube.

TASKS
Rubik's Cube has an astronomical number of ways it can be permuted, and the code had to be efficient with time and memory usage. I used C++ due to its high level of performance and control with which the programmer can manage the memory. The path-finding algorithm known as A*, i.e., A Star, was one of the main actors of this solution and worked especially well with this kind of problem.

After code optimizations, tests, and refactoring, a robustly designed program was devised that could solve any shuffled instance of Rubik's cube using an acceptable amount of time and memory.

The client requested multiple loan calculators that would compute metrics, such as the repayment plan, interest cost, and amortization schedule of conventional amortized loans.

As a full-stack developer, I created these calculators and presented them in a user-friendly UI within the web app.

A neural network system was developed using MATLAB with the purpose of imitating a successful stock investor.

The system would be fed with the trading patterns of such investor and back-tested to ensure consistency. It was also tested against real-time forward-looking data.

A book was then published following its success. A link to the book is shown in the project URL.

My main contribution to the Life-Assist company so far has been a complete redesign of its web app's homepage. Their marketing team documented all the design requirements for a successful redesign. Completing this project required implementing CSS animations, dynamic menus, carousel sliders, JavaScript and TypeScript functions, and more. A responsive design and compatibility with major browsers were critical requirements. Back-end work was also essential because the homepage is closely connected to many modules within the API. Therefore, I ended up working with .NET, C#, and SQL Server to implement new modules and controllers while extending the API's core functionality. New SQL tables and stored procedures were also created. The result was a fully modernized, enhanced, and more stable homepage where the IT coordinator responsible for this project exposed his total satisfaction with the new homepage.

An automated trading bot based on C# developed using Interactive Brokers APIs. This would trade within the Forex markets following specific technical analysis rules, exploiting certain market deviations. The bot was previously backtested using historical market data.

Ngöndro Tracker is an offline-first mobile application built with React Native (Expo) and TypeScript that helps practitioners track progress in diverse meditation practices. The app uses a local-first architecture with SQLite for persistence, allowing users to log meditation sessions, monitor cumulative counts, and track progress toward long-term repetition goals even without network connectivity.

Each entry is modeled as a session, with totals derived dynamically to ensure data integrity and auditability. The application includes practice management, streak tracking, daily analytics with chart and table visualizations, and a backup import and export system. Optional user authentication and cloud synchronization are implemented using Supabase, enabling secure account access and cross-device data portability while preserving the app’s offline-first design.