These projects show different ways of solving problems: integrating services, replacing legacy systems, responding to unexpected changes, and creating tools to work with AI. In each case I explain my involvement, the main decisions, and the status of the solution.
Case studies
In operation and evolution.
Pipila
Pipila was born to resolve an urgent migration and ended up becoming a reusable integration solution. I built a modular API that separates the interface from processing, preserves pending records in case of external failures, and allows incorporating new flows without depending on the previous platform.
I created Onix to replace a system based on Clipper that hindered work over the network and required frequent data corrections. The ERP continues to handle the corporate operations. I am currently participating in its modernization, with the code migrated to Laminas and a test environment on AWS.
GUACAMAYA was born from a personal challenge: developing a tool that would help me write and publish a book before a convention, with less than a month available. It continues to evolve as I use it in my editorial work, keeping control over the ideas and the outcome.
Historical case; current operational situation not confirmed.
Winner
Winner emerged when the withdrawal of an external platform put a near-launch campaign at risk. I reused components, added a Laravel service and deployed the solution on AWS. The development allowed meeting the deadline and was later reused in other campaigns.
Used in real work; maintenance cycle not specified.
Yayauhqui
I developed Yayauhqui with AI assistance to evaluate evidence of AWS infrastructure and identify possible security vulnerabilities. Its operation uses heuristic algorithms: it does not incorporate an AI model to perform the analysis. I have applied it to the review of infrastructure in a real work context.
Used on a website; maintenance cycle not specified.
SpellChecker
SpellChecker uses a trained model and LocalAI to check spelling on websites. I used it to validate a website I developed. It is an example of applying AI to a specific need of content review in a language other than my own.
In development, with application in a real work methodology.
LORO
LORO is a memory edge layer that I developed to expand and improve the context of work with agents. It arose from my experience with complex creative tasks and the need to reconstruct information between stages. It has been used in the modernization of Onix as part of an AI-assisted development methodology.
In development; no confirmed commercial offer available.
PERICO
PERICO is an extension of VS Code that implements LORO to automate functions related to memory and agent context in software development. It is still in development and its repository is private. My intention is to offer it as a service.
Delta combines backend and frontend components, REST services and WebSocket, databases and integration and deployment processes on AWS. It is part of my experience creating reusable bases for different solutions.
Delta Commerce derives from Delta and applies a modular and headless architecture to e-commerce, with integration with Shopify. Its approach separates the user experience from the services that support the operation.
K4Y is a platform oriented toward loan, capital, interest, and SPEI transfer management. It is part of my experience in financial process solutions and in integrating specialized functions within an operation.
Holstein is a network oriented toward breeders, producers, and buyers of livestock. The project is part of my experience developing solutions for sectors with specific needs and forms of interaction.