Ignacio – Python, React, LLM

A pharmaceutical company is focused on integrating assistants to help with different communication channels so that the user can access various tools from multiple sources.

• Implemented workflows of conversation using Langgraph and Langchain;
• Integrated flows of conversation into already working platforms.

A management platform for Drone flights includes HD Maps, flight routes, live alerts, and everything related to drone flight operations.

• Created the whole system from zero;
• Built real-time notifications;
• Implemented custom KMZ processing system with Container Apps dynamic scaling.

A consulting company that provides various developments focused on LLM integration. The main projects were Multi-Agent Assistants with RAG integration to provide access to technical information through a Chat UI integrated with SharePoint.

• Deployed systems to Serverless, Container Apps, and Kubernetes environments;
• Designed Langgraph systems with Python and Typescript;
• Implemented a complete stack system from 0 with NextJS and Python;
• Managed SharePoint connections with the RAG and the frontend for permission-based file access.

A web platform allowed users to be diagnosed by a Dialog Flow and a Vision Machine Learning model that reviewed dental RXs from the patient to detect early caries and anomalies. The project was funded by government funds and was shut down after a few years of operation.

• Designed the software for the platform;
• Implemented the FastAPI backend and the React frontend;
• Deployed the application into a Docker Swarm node;
• Executed planning and task assignments for 1 frontend developer and 1 backend developer;
• Trained a MaskRCNN model for detecting caries and anomalies in RXs.

A company that implements automation and machine learning solutions. Cooperation involved 2 main projects:

• Implementation of a vision model for detecting personal security element alerts for workplaces based on analyzing images sent through Kafka.
• Implementation of 3 different Machine Learning models for detecting caries and anomalies in dental RXs.

• Trained YOLO vision models on custom data with results of more than 85% accuracy;
• Implemented automation services in EC2 for the YOLO vision execution using rq queues;
• Worked with MaskRCNN vision model on custom data with GCP GPU clusters to achieve between 75 and 85% accuracy;
• Trained a U-Net model for the detection of dental pieces in RXs.