A. Rosa Castillo, PhD

rchitected production-grade Agentic AI systems (LangGraph), optimizing the multi-step reasoning loops to maximize accuracy while minimizing computational overhead and energy intensity. Implemented a spam detector for security reports, achieving 70% accuracy in detection. Developed rigorous performance and sustainability metrics for LLM agents, establishing the empirical baseline for evaluating the 'Green AI' trade-offs of autonomous security workflows

Prepared technical interview with different technical exercises. Completed full interviews with several candidates for the role requested by the client. Created insightful reports about the candidate's performance on the interview.

Optimized autonomous agent frameworks (CrewAI) for blockchain data, focusing on inference efficiency and hardware-aware algorithm design to reduce the environmental footprint of real-time forecasting. Improved the current trading algorithm used for the blockchain agents. Benchmarked and integrated high-performance LLMs, evaluating model selection through the lens of carbon intensity vs. predictive power.

Helped the research team—committed to building a high-quality analysis package—incorporate some best practices regarding software architecture, coding guidelines, and testing. Contributed to the improvements introduced in the documentation about the Python package, making the code more robust, reliable, scalable, and easier to maintain. Instructed the small team on the next steps and how to follow the best coding guidelines as part of the engagement.

Collected Blockchain data and performed accounting algorithms on it, expanding my knowledge of Blockchain technology. Explored finance terminology and how to build meaningful reports for stakeholders. Started using Neo4j to implement graph analysis and represent data more efficiently than just applying plain relational database models.

Built more efficient SQL queries to deal with a huge data volume and created my first reports in Tableau. Created data pipelines using Kafka streaming and learned to use Docker for effective testing. Worked on my very first project in Spark to compute fraud detection metrics.

Performed a deep customer service analysis based on ticketing data for various companies. Developed different NLP techniques as part of the data product offered to clients. Improved reporting practices by sharing visualization graphs with the clients.

Managed a team of two for three months while looking for a new project manager. I also wrote risk analysis for a project, conducted computer estimations, and handled employee performance reviews. Handled requirements engineering carefully, considering regulations and tracking needs in the healthcare sector. Became certified in software requirements engineering by the International Requirements Engineering Board. Worked on a refactoring project with a challenging new system's software design and architecture. I learned a lot about software patterns and algorithm optimizations.

Worked on the system architecture, analysis, and design of software components using UML, Enterprise Architect, and other software engineering tools. Embedded code in a Linux-based device used in trains to manage train announcements. Coded with C++ and handled testing both in the lab and running trains.