Training, fine-tuning, and deploying large language models at scale. Focused on robust evaluation, data-centric AI, and production-grade LLM platforms with real user impact.
Based in Buenos Aires, Argentina
Artificial Intelligence Engineering student at the University of San Andrés (UdeSA) and NLP Engineer.
I train, fine-tune, and distill large language models with direct impact on real products and users.
My work focuses on scalable data architectures, rigorous evaluation frameworks, and end-to-end LLM training and serving platforms.
NLP Engineer – Mercado Libre (Jan 2026–Present)
- Train, fine-tune, and distill large language models with direct impact on Mercado Libre products and millions of users.
- Design data architectures, metrics, and evaluation frameworks to ensure robust, scalable, and high-quality LLMs.
- Build and continuously evolve large-scale AI model training and serving platforms.
Senior Data Scientist – SOFLEX (2025–Jan 2026)
- Trained and fine-tuned language models for large-scale incident and security reporting systems.
- Designed data pipelines, metrics, and evaluation loops for models operating over 200M+ textual records.
- Built and deployed NLP services with low-latency inference and continuous monitoring.
Research Assistant – LiNAR, UdeSA (2025–Present)
- Researching representation learning and self-supervised objectives, with emphasis on scalable training and evaluation methodologies.
AI Consultant and Lead – Papelera San Andrés de Giles (2025–Present)
- Designed multimodal pipelines combining OCR, embeddings, and language models for industrial-scale classification.
- Productionized AI systems with robust serving, monitoring, and retraining strategies.
AI and ML Lead – UK University (2025)
- Built and deployed NLP systems for intent detection and large-scale lead processing.
- Accelerated model iteration cycles by introducing standardized training and evaluation frameworks.
AI Engineer – Notimation (2023–2024)
- Implemented RAG, agentic workflows, and LLM serving architectures in production environments.
- Improved reliability and scalability of NLP systems through better data curation and evaluation.
- Languages: Python, C/C++, Java, JavaScript, TypeScript, CUDA
- NLP & LLMs: Transformer architectures, fine-tuning, distillation, RAG, evaluation & benchmarking
- Machine Learning: PyTorch, TensorFlow, PyTorch Geometric, LangChain
- Infrastructure: Docker, Kubernetes, AWS, GCP, SQL, DynamoDB, Cassandra
- B.Sc. Artificial Intelligence Engineering – UdeSA (2022–2027)
- 95% merit scholarship. GPA 8.67/10
LinkedIn: linkedin.com/in/lebrero-juan-francisco
GitHub: github.com/frizynn