Robert – Python, AI agent development, LLM

An end-to-end outreach automation platform aimed at streamlining and managing communication workflows efficiently. The platform allowed users to automate outreach processes, track responses, and optimize engagement strategies. The system was built for cost efficiency, deployed on a small virtual machine, and relied entirely on open-source tools and in-house solutions.

• Designed and implemented the full backend architecture, including database schema, API schema, and routes;
• Built most of the API logic and integrated third-party webhooks for event-driven actions;
• Developed an agentic system for LinkedIn outreach automation and integrated it with LinkedIn via third-party APIs;
• Connected the platform with various LLM providers to enhance automation capabilities;
• Implemented unit and integration testing to ensure code quality and reliability;
• Set up monitoring using the Grafana stack to track system performance;
• Configured CI/CD pipelines with GitLab CI/CD, hosted on the virtual machine, for seamless deployment.

An agentic AI software solution using LangGraph and Azure OpenAI that automated document analysis for commercial certification applications in Mexico. It was deployed as a backend API through fastAPI, Docker and Azure Web Services. It reduced the processing time of certification applications by over 80% and enhanced employee productivity within 6 months.

• Planned and implemented an agentic AI solution;
• Designed and built the backend architecture;
• Implemented unit and integration testing to ensure code reliability;
• Set up monitoring using the Grafana stack;
• Configured CI/CD pipelines using CircleCI for automated deployment.

The entire backend for a cybersecurity certification app, streamlining risk assessment with a multi-step analysis process and a GraphRAG chatbot. Deployed the API using FastAPI, Docker and Azure Web App.

• Designed and implemented a scalable backend architecture to support web application functionality;
• Deployed the application on Azure, ensuring efficient resource usage and reliable hosting;
• Developed comprehensive unit and integration tests to maintain high code quality and stability;
• Set up monitoring using the Grafana stack to track system performance and detect issues proactively;
• Automated deployment and test execution workflows using CircleCI, streamlining the release process;
• Built a GraphRAG chatbot for the website using the LightRAG library, enabling interactive user support and content retrieval.

A product recommendation system using PyTorch to improve personalized suggestions for users. The system optimized targeted email campaigns, increasing user engagement and click-through rates by 20% over six months. The project involved designing the recommendation algorithm, implementing data preprocessing, and integrating the model into existing marketing workflows.

• Developed a recommendation system that significantly outperformed the company’s previous solution;
• Defined and implemented metrics for measuring and monitoring model performance;
• Designed and executed data preprocessing pipelines to ensure high-quality input for the model;
• Wrote clear documentation to support maintenance and knowledge sharing;
• Implemented CI/CD pipelines using Azure DevOps to automate deployment and testing workflows.