Sara Ana

Building a web-based Student Career Advisor System in collaboration with IBM. The system matches student profiles to job opportunities using structured skill data and provides AI supported, human readable explanations for each match. Working as full stack, including a frontend for user interaction, backend services for business logic, and persistent storage using PostgreSQL and MongoDB for user, profile, and job data. Integrating an AI component based on LLMs with a RAG pipeline to explain matches in natural language. The work will emphasize keeping the system simple, explicit, and easy to extend.

Worked on building a forecasting model for the Gemeente of Amsterdam.

Worked on building and evaluating ML forecasting models for short-term natural gas prices, in the scope of optimising market driven decisions. Delivered model suites(AutoRegressiveIntegratedMovingAverage, Hybrid Gaussian Process, GRU and LongShort Term Memory) that forecast TTF-related prices across multiple horizons with probabilistic outputs and a decision-focused dashboard.

Worked across the full stack with .NET technologies, contributing to both front-end and back-end development. Used C#, TypeScript, HTML, and CSS to build and maintain scalable features, mainly supporting CPQ-related functionality. Handled database operations, including SQL queries and migrations, and created GitHub workflows to automate pull request processes. Actively participated in code reviews and team discussions to maintain high coding standards. Took over and completed a roles and permissions project designed to embed a consistent authorisation layer across the application, finalising it for a stable 2.0 release. Throughout the project, maintained strong use of Entity Framework to ensure consistent data access patterns and business logic across the application. At the moment, researching and validating AI technologies, as well as improving development with AI tooling (LLM exception solver, vector databases for retrieval augmented)

As Sublead for the Software Maintenance team, I develop and support the robotic arm simulation for the European Rover Contest (ERC). I work mainly with C++, Python, and Linux-based systems, building simulations in RViz and Gazebo by assembling URDF/Xacro models and setting up motion planning through MoveIt. I implemented keyboard teleoperation to enable manual control for testing and debugging. My responsibilities also include: - creating and managing launch files to coordinate multi-node project deployment - maintaining a clean project structure - improving simulation stability and performance. In addition to development work, I mentored new team members, aiding in onboarding, and helped organise the team's tasks to ensure steady progress.