rubencr.com
Sign in

AI Engineer & Researcher

Building intelligent systems from first principles to real-world impact

I combine AI, computational science, physics and mathematics to understand complex systems — and turn that understanding into useful technology.

Get in touch

  • AI Systems

    From classical ML to LLMs

  • Computational Biology

    Simulating molecular systems

  • Physics & Mathematics

    Understanding first principles

  • Real-World Impact

    Turning ideas into systems

Rubén Cañadas, arms folded, against a dark field of amber and violet light trails

About me

Hi, I'm Rubén Cañadas

AI Engineer & Researcher

I build intelligent systems that turn into real products and real businesses.

I come from physics and mathematics, and I work across the whole path an AI system has to travel: the research that makes it possible, the engineering that makes it fast and reliable, and the product decisions that make it something people actually pay for.

Research only matters to me once it leaves the lab. I care about the systems that move a real number — revenue, margin, time to market — not the ones that only move a benchmark.

  • Performance
  • Alignment
  • Systems
  • Research
View my work

Focused on what matters

I work on the problems that decide whether an AI system becomes a product or stays a demo: performance, scalability and alignment, in that order of stubbornness.

Explore my research

  • 10+

    Years of experience

    in AI engineering and research

  • 5+

    Systems shipped

    from concept to production

  • 2

    Scientific publications

    peer-reviewed research

  • Built to ship

    Tools and research that end up in products people pay for

Expertise

Three areas where I create impact.

At the intersection of AI, computational science, and fundamental theory, I build systems that generate real-world value and expand our understanding of how the world works.

AI Engineering

Practical AI that delivers value.

I design and build AI systems that solve real problems. I start simple and iterate: if a Random Forest or SVM gets the job done, that's the tool. When generative models are the right choice, I fine-tune and adapt small language models and combine them with retrieval and structured data to deliver accurate, reliable, and scalable solutions.

  • Fine-tuning
  • Small Language Models
  • RAG & Knowledge Graphs
  • Embeddings
  • Evaluation
  • ML (RF, SVM, XGBoost)
A glowing hexagonal model core connected by luminous threads to icons for documents, data, tuning and evaluation

From data to decisions: combining models, retrieval, and tools to build AI systems that are useful, trustworthy, and efficient.

Computational Biology

Simulating life at the molecular level.

I use physics and mathematics to model and predict how biomolecules behave in realistic environments. From molecular dynamics and Monte Carlo simulations to docking and free energy calculations, I build computational experiments that accelerate discovery and reduce uncertainty.

  • Molecular Dynamics
  • Monte Carlo
  • Molecular Docking
  • Free Energy Methods
  • Enhanced Sampling
  • Biophysics
A protein ribbon structure in violet and amber floating above waves of glowing particles

Turning physical laws into insights about proteins, ligands, and cells — in silico experiments that complement wet lab.

Physics & Mathematics (Hobby)

Understanding the universe through first principles.

Outside of applied work, I explore the foundations. Differential geometry helps me understand how spacetime is modeled as a curved manifold. Linear algebra and optimization on manifolds, like the Grassmann manifold, serve as the language to describe quantum states and optimize orbitals. It's curiosity-driven, but it deepens how I think about complex systems.

  • Differential Geometry
  • General Relativity
  • Quantum Mechanics
  • Grassmann Manifolds
  • Optimization
  • Theoretical Physics
A spiral galaxy above a curved spacetime grid, next to violet planes representing manifolds

Exploring the mathematical structures behind reality — from curved spacetime to quantum manifolds.

Let's build something meaningful.

Open to collaborations and challenging problems.

Get in touch

Trajectory

From computational biology to production AI systems.

A decade-long journey across research, machine learning, and AI systems — turning science into real-world impact.

2016 – 2019

Computational Biology

Modeling biological systems

Model biological systems using molecular dynamics, Monte Carlo, quantum chemistry, enzyme engineering, and virtual screening.

  • Molecular Dynamics
  • Monte Carlo
  • Quantum Chemistry
  • Enzyme Engineering
  • Virtual Screening
  • ML for Biology
2020 – 2021

Applied Machine Learning

Forecasting & data science

Build forecasting models, engineer features, apply deep learning for time series, and turn data into business insights.

  • Forecasting
  • LSTM
  • CNN
  • Feature Engineering
  • Data Pipelines
  • Business Insights
2022 – 2024

Production ML

Scaling models into systems

Operationalize machine learning with MLOps, supervised learning, reinforcement learning, monitoring, and production systems.

  • MLOps
  • Supervised Learning
  • Reinforcement Learning
  • Time-Series
  • Model Monitoring
2024 – Present

AI for Drug Discovery

Scientist-facing ML platforms

Build end-to-end ML systems with protein models, cloud infrastructure, deployability/PTM models, and tools used by scientists.

  • End-to-End ML Systems
  • Protein Models
  • AWS
  • EKS
  • ArgoCD
  • Scientist Tools
  • Drug Discovery
Current Focus • 2026

AI Systems

Building robust AI systems.

Advance alignment, CUDA kernels, LLM infrastructure, and research that leads to product impact.

  • Alignment
  • CUDA Kernels
  • LLM Infrastructure
  • Performance
  • Research to Product Impact

Selected work

Things I build and maintain.

Open source

MolFun

Protein structure fine-tuning

Open-source framework for adapting pre-trained protein structure prediction models, with four fine-tuning strategies and a modular registry.

  • PyTorch
  • CUDA
  • Triton

Nomosis

Molecular intelligence platform

AI agents, molecular dynamics and drug discovery in one ecosystem: pose and trajectory viewers, research notebooks, HPC orchestration.

  • AI Agents
  • MD
  • HPC
Private

CodeMol

Console-first protein viewer

Desktop 3D molecular rendering driven by a terminal, with 200+ tools and LLM agent routing.

  • Python
  • WebSocket
Open source

Agent Skills

Skills for Claude Code

Production-tested skills encoding best practices for backend, web, testing and security work.

  • TypeScript
  • MCP
Launching May 2026

Novachef

AI cooking assistant

Mobile app with 3000+ recipes, a RAG-backed assistant, shopping lists and menu planning.

  • RAG
  • Mobile

Writing

Notes from the work in progress.

All articles
Thoughts

Can we explain most of the world with less information?

A reflection on information, perception, AI, molecular dynamics, and whether complex systems can be understood through much smaller latent representations.

Machine Learning

Positional Encoding in Transformers: From Sinusoidal to RoPE

Understand positional encoding in Transformers, from sinusoidal embeddings to RoPE, and learn how models capture token order and relative position.

Machine Learning

PyTorch Compiler Explained: TorchDynamo, AOTAutograd & TorchInductor

Understand how PyTorch compilation works under the hood: bytecode capture with TorchDynamo, forward/backward staging with AOTAutograd, and kernel optimization with TorchInductor.

Let's connect

Interested in discussing research, ideas or collaborations?

Always open to connecting with like-minded researchers, engineers and innovators.

rubencr14@gmail.com

  • Based in

    Europe

  • Time zone

    CET (UTC+1)

  • Available for

    Research collaborations

    Technical discussions

    AI alignment projects