TLDR: Senior AI Engineer, 4+ years of experience driving cloud-based AI/ML products to market in the AI for Science, Self Driving, and Entreprise Data spaces, skills: Python, Rust, C#, TypeScript, PyTorch, Tensorflow, AWS, GCP, Docker, SQL, REST, websockets, French, Spanish. Learn more at federicoarenas.io.
Senior AI Engineer (11/2025 → Present), AI Engineer (04/2024 → 11/2025) — materiom.org
Funded by Google, end-to-end development of production-ready AI software for bio-based materials, focusing on generative AI, LLMs, integrating tools and knowledge bases from materials science and biochemistry:
- Materiom AI: Agentic RAG application with access to 40000+ materials and various search APIs, serving 10000+ user sessions since its deployment, and 10+ entreprise partnerships.
- Materiom Commons: web-based platform accelerating bio-based material innovation, catering to 25000+ registered users across 130 countries with 99.99% uptime.
- Neural models for material synthesis and material property prediction trained on experimental data from partner scientific labs across the world. Trialed with multiple entreprise partners.
- Evaluation pipelines for automated, continuous model monitoring in production, including expert-led metrics to enable rapid diagnostics and model improvement.
- Engaged in key investor and partner events to expand our customer base and establish strategic partnerships in Europe, the US, and the UK.
Stack: GCP; Cloud Run, PGVector, PostgreSQL, MongoDB, FastAPI, Python, ADK, Langfuse, PyTorch, React, TypeScript.
AI Engineer (09/2021 → 04/2024) — conode.ai
Led the design, development, and deployment of cloud-based AI-human collaboration features at Conode (previously dRISK):
- Enterprise Data Solution: web app built with TypeScript, Rust, and Python, enabling non-technical decision-makers to navigate and understand large datasets graphically.
- AV Perception Retraining Platform: Allows users to graphically identify and fix perception failures through data curation, model retraining, fine-tuning, and performance tracking.
- LLM-Powered Agent: Enables non-technical users to interact with data exploration tools via natural language, hosted on a Python Flask server with websocket communication.
- Led multiple customer-facing projects from PoC to mature relationships with major clients like NVIDIA, Luminar, TATA Motors, SAE and others.
- Engaged in key investor events to expand our customer base and form strategic partnerships, especially in SE Asia and the UK.
Stack: Rust, PyTorch, Flask, Python, React, TypeScript.
ML Engineer (06/2021 → 09/2021) — neurolabs.ai
Led the exploration for learned Synthetic Data Generation methods integrated into their Retail Product. Resulted in a Distinction-grade thesis. Read more here.
Education
MSc in Artificial Intelligence (The University of Edinburgh, grade: distinction), MSc in Mechatronic Engineering (Arts et Métiers ParisTech, grade: 17/20), BSc in Product Design Engineering (EAFIT University, grade: 4.75/5.0).
Relevant modules: Calculus, Linear Algebra, Optimization, Accelerated NLP, Machine Learning Practical, Reinforcement Learning, Natural Language Understanding, Generation and Translation, Image and Vision Computing.
Patents
(Application No. PCT/US2024/032276): ML algorithm using graph theory and A* search to augment trajectory data for training autonomous vehicle RL models.
Awards
Colfuturo Excellence Scholarship, Eiffel Excellence Scholarship.