Matías Castro

Engineered complex prompts in English and Spanish for LLM training and evaluated LLM responses across multiple dimensions. Annotated audio and video datasets to support AI model development and reviewed peer submissions for consistency and quality.

Annotated diverse datasets including image, audio, video, and text using Labelbox. Contributed to AI model training through RLHF and SFT techniques, including LLM fine-tuning, while ensuring high-quality annotations by adhering to strict guidelines.

Defined and documented comprehensive QA processes, including regression testing and metrics tracking. Successfully migrated and reorganized test cases using Zephyr, and implemented Gherkin for efficient test case design and peer review processes.

Performed ambiguity analysis on requirement documents to identify potential issues early in the development process. Designed and executed detailed test cases for marketing applications, and reported defects with clear documentation for efficient resolution by developers.

Designed and executed test cases for both web and mobile applications, effectively reporting defects and tracking resolutions using Jira. Contributed significantly to defining user stories and suggesting product improvements.

Led a team of nine testers on Facebook game projects, including one that won Best Facebook Game of the Year in 2010. Implemented a night shift for three months to meet critical deadlines and mentored team members to improve performance and resolve conflicts.

Managed a team of seven testers, coordinating effectively between developers, testers, and producers to streamline workflows. Conducted thorough tester evaluations and contributed significantly to strategic decision-making processes for the QA area.

Was one of the first three members to start the Córdoba QA team, actively contributing to testing and quality assurance processes for game development. Played a foundational role in establishing initial QA operations.

Actively participated in comprehensive testing and quality assurance processes specifically for game development projects. Contributed to ensuring the high quality and functionality of various gaming titles.

Developed a testing system for an AI-driven object recognition platform, creating video-based test footage using VFX techniques to overcome real-world constraints. Refined the AI model by annotating hundreds of hours of video footage, ensuring high-quality training data, and collaborated with the data science team to optimize model performance.

Currently pursuing a degree in Film and TV. This program focuses on the theoretical and practical aspects of cinematic and television production.