Darío Cabezas Larese
Understanding
intelligence.
Making it useful.
My work connects AI research, industry and product development. I study how models learn, develop AI systems for practical problems, and build software around everyday needs.
Currently, across
three connected worlds.
01 / SELECTED WORK
Original research figure · Yang et al., 2026FEATURED RESEARCH · NEURIPS 2026
Learning how
attention sees.
What if a transformer could learn a richer notion of similarity? Our work explores nonstationary attention kernels, shaped by normalizing flows.
Flow-based Spectral Kernel Learning for Nonstationary Attention.
Explore the research ↗Watch my explanation ▶AI beyond
the prototype.
Classification and entity resolution for the complexity of enterprise data.
Explore my industry work ↗INDEPENDENT PRODUCT / iOS
Quickie.Finance appA little less friction.
A lot more clarity.
An independently built iOS app for tracking everyday income and expenses.
Meet Quickie ↗02 / ONE THREAD, THREE PERSPECTIVES
From understanding
to application.
Investigate.
I study transformer architectures, attention mechanisms and probabilistic modeling for sequential data, including text and time series. The starting point is a question; the standard is an experiment.
↗Apply.
I turn complex organizational data into useful AI systems, with evaluation and engineering close to the model.
↗Build.
I make independent products to explore the whole journey: from an everyday problem to an interface someone can use.
↗RAILWAY VIBRATION FORECASTING / IEEE · 2025
What can a railway
tell us through vibration?
Changing speeds and loads make axle vibrations difficult to predict. In this study, I investigated how transformer models could help forecast these signals.
Read the research story ↗A SMALL EXPLORATION
The past changes
what comes next.
In a self-exciting Hawkes process, each observed event raises the modeled rate of future events, and its influence decays over time. My research applied this framework to the timing of recurrent suicide attempts. Explore the mathematical idea below, or read the research.
Illustrative exponential Hawkes process. Synthetic events, arbitrary time units; not patient data or a clinical prediction.
The mathematical idea
λ(t) = μ + Σ α exp(−β(t − tᵢ)), summed over tᵢ < t. Here μ = 0.3, α = 0.7 and β = 0.8; α/β < 1. The curve shows the conditional rate given the selected event history, not an event probability.
03 / EXPERIENCE & AFFILIATIONS
A broader perspective.
Across disciplines.
Working across AI research, industry and independent software development gives me a broad perspective on how ideas become useful systems. I bring questions from applied work into research, and use what I learn about attention mechanisms, probabilistic modeling and sequential data to inform practical AI problems. Building products adds another perspective: connecting technical decisions with usability. This exchange across disciplines shapes how I approach problems, evaluate methods and build software.
A little more about me ↗
LET’S WORK TOGETHER
Have an
interesting problem?
Bring your AI challenge, research question or product idea. Let’s define a useful direction and the next steps to make it happen.
Discuss your project ↗