Open to collaboration · Santiago, CL
AI Engineer · Backend · Data · Education
I design, build and operate applied AI systems for education and data governance. I work with LLMs, RAG, embeddings and production-oriented backend architectures.
erickfmm@gmail.com +56 9 6183 0057
Frankenstein Transformer does not stop at the paper: the experiment atlas and the configuration builder are open to explore.
2.66 M training steps, 56 figures, crossed analysis and downloadable datasets. The question: which mixer pair learns fastest at 32 tokens.
Open the atlasYAML validated against the schema, parameter estimation and the training command ready to copy.
Open the builderI combine research, backend development and data governance to build useful, auditable systems that are ready to operate.
Computer Science Engineer with a focus on database design. I work across project management, AI-assisted backend development and data analysis with Python.
Experience in requirements analysis, language-model proofs of concept and data governance for the public sector.
Languages: Python (advanced), C# (basic), JavaScript (basic)
Data: PostgreSQL, MongoDB, Qdrant
Backend & ML: Django, FastAPI, Flask, Scikit-Learn, SciPy, LangChain, Transformers, Sentence-Transformers
Tools: Docker, Tableau, Google Looker, Power BI, OpenMetadata, Gephi
AI, data governance and education: from requirements analysis to operating systems.
Innovation Center Professional
Research Assistant
Intern (2nd internship)
STEM Assistant
Intern (1st internship)
Formal training in engineering, databases and data science.
Universidad de Valparaíso · 2023 · 6 months
Passed with Highest Distinction (6.1).
Universidad de Valparaíso · 2019
Specialization in Database Management and Design. Graduated with Highest Distinction (6.22).
Open-source tools, automation and applied research.
A config-driven Python library and CLI for training experimental Transformer architectures. It supports Mixture of Experts, BitNet 1.58, Titan Memory and more through YAML/JSON configs.
A Frankenstein Transformer plugin for DashAI that brings its experimental architectures into the DashAI ecosystem.
Metaheuristics and numerical optimization library with advanced visualizations: PSO, DE, CMA-ES, Black Hole and more.
A curated list of resources, datasets, models and courses for NLP and LLMs in Spanish.
Converts Markdown books into multi-chapter audiobooks with Spanish TTS engines.
Scripts to translate, filter and prepare datasets for publication on Hugging Face.
Automatically generates educational presentations from YAML into PPTX and PDF.
Research in multimodal learning, educational analytics and EEG-based emotion recognition.
A new EEG software that supports emotion recognition by using an autonomous approach
Neural Computing and ApplicationsExploring Collaborative Writing of User Stories with Multimodal Learning Analytics
IEEE AccessDevelopment of a Software that Supports Multimodal Learning Analytics: A Case Study on Oral Presentations
Journal of Universal Computer ScienceMathematics Learning through Computational Thinking Activities: A Systematic Literature Review
Journal of Universal Computer ScienceUsing Black Hole Algorithm to Improve EEG-Based Emotion Recognition
Computational Intelligence and NeuroscienceEvolução das posturas corporais de estudantes em apresentações de seminários
Brazilian Symposium on Computers in EducationUsing Multimodal Data to Find Patterns in Student Presentations
LACLO 2018Developing a Dashboard to Support the Analysis of Multimodal Educational Data
Congreso Chileno de TICS para la EducaciónSpanish-language models and datasets for NLP, embeddings and experimentation.
Explore more collections, models and datasets on the Hugging Face profile.
View profileIf you have an AI, data or education project in mind, get in touch.
Send me a short project summary, the context and the outcome you are looking for. I reply by email.
