Eduards A. Chipatecua

I am interested in understanding how computational systems work, but even more in identifying questions whose answers are not yet clear to me.

I am a Computer Scientist from the National University of Colombia and currently an M.Sc. student in Computer Science at the University of the Andes. My academic interests span artificial intelligence, machine learning, algorithms, programming languages, scientific computing, and computer systems.

I am particularly drawn to problems that appear at the intersection of these areas. Questions about machine learning can quickly become questions about numerical precision, optimization, compilers, concurrency, hardware reliability, or energy consumption. I enjoy following these connections because they often reveal that what initially looks like a problem in one area is actually part of a much broader computational question.

A question posed by Richard Hamming has strongly influenced the way I think about research:

“What are the important problems in your field, and why aren’t you working on them?”

I find this question valuable because it forces me to think not only about whether a problem is technically interesting, but also about why it is worth investigating. At the same time, I do not believe that meaningful research must always begin with a grand challenge. Interesting problems often emerge from much smaller observations: an unexplained experimental result, an inefficient computation, an unexpected limitation of an algorithm, or simply something that does not behave as expected.

For this reason, I maintain a collection of open questions that I find mathematically, computationally, or scientifically interesting. Some may eventually become research projects; others are simply questions that I would like to understand more deeply.

My current interests include continual learning, deep learning, scientific computing, program analysis, compiler optimization, numerical efficiency, and computational systems. What interests me most is not only whether a computational method works, but why it works, under which assumptions it works, when it fails, and what ultimately limits it.

During my undergraduate studies, I had the opportunity to work under the supervision of Professor Fabio González at MindLab. One of the experiences that most influenced my understanding of research was contributing to the recovery and reopening of the laboratory’s computing infrastructure. I worked with Linux servers, CUDA-enabled GPUs, monitoring tools, technical documentation, and access to computational resources for graduate students and researchers.

That experience changed the way I think about scientific work. Research depends not only on ideas and algorithms, but also on the infrastructure that makes experimentation possible. Reliable systems, reproducible environments, computational resources, documentation, and shared technical knowledge can determine which questions a research group is actually able to pursue.

I currently collaborate with Professor Juan Galvis and Professor Francisco Gómez on initiatives related to scientific and high-performance computing at DataLab.

One question that has recently attracted my attention is the relationship between numerical precision, computational performance, and energy consumption. I have worked on experiments studying how different numerical representations affect execution time and energy usage in scientific computations. These problems have strengthened my interest in the relationship between algorithms and the physical machines that execute them.

Computation is abstract when we describe it mathematically, but its execution is never abstract. It has concrete costs in time, memory, communication, energy, and hardware utilization. Understanding those costs—and how they influence the algorithms and models we design—is something I would like to explore further.

My way of learning is strongly based on reading, implementation, experimentation, and problem solving. I prefer to study technical material slowly enough to reconstruct arguments, work through examples, and identify assumptions that I do not fully understand. Implementation is especially useful to me because it provides a direct test of whether my conceptual understanding is actually correct.

I keep a reading page with books that I am currently studying, books that I have worked through substantially, and material that I would like to study in the future. I also read research papers and often build small computational experiments around questions that arise from them. These experiments are frequently exploratory: the goal is not necessarily to obtain a final result, but to understand the structure of a problem well enough to ask better questions.

Outside formal coursework and research, I enjoy mathematical and algorithmic problem solving. I regularly work on problems from Project Euler, where relatively compact problems can lead naturally to number theory, algorithms, computational complexity, numerical methods, or questions about how to search a solution space efficiently.

For me, obtaining a correct answer is usually only the beginning. I am often more interested in understanding why a method works, whether the argument can be generalized, whether a simpler formulation exists, and whether the computation can be made more efficient.

Teaching and mentorship are also an important part of the academic work I would like to do. At the National University of Colombia, I have worked with undergraduate students, particularly students beginning their studies in computer science. I helped develop a student mentorship community intended to provide academic guidance, share opportunities, and encourage students to become progressively more independent learners.

What I value most in mentoring is not simply helping someone solve an exercise. I am interested in helping students become capable of formulating questions, locating resources, reasoning independently, and continuing to work when a problem is initially unclear.

I also believe that academic communities are part of the infrastructure of research. Knowledge, opportunities, computational resources, and experience become considerably more valuable when they circulate among students and researchers. In the long term, I would like research and teaching to remain closely connected parts of my academic work.

I am also interested in how science is communicated. I founded Todos hacemos Ciencia, a science communication initiative centered on conversations with students, researchers, and academics about research, higher education, scientific practice, and the experiences behind academic work.

I am particularly interested in communicating the process of science, rather than presenting research only through polished final results. How did someone arrive at a question? Which approaches failed? Why did a problem become interesting? How did an idea change during the research process? I think these questions often reveal as much about scientific practice as the final publication itself.

I do not expect my academic interests to remain fixed. At present, I am especially interested in learning systems, algorithms, programming languages, scientific computing, and efficient computation, but I value being able to follow questions that lead beyond the boundaries of those areas.

Some questions begin as research problems. Others emerge from a mathematical puzzle, a textbook, an unexpected program behavior, or an experimental result that does not agree with intuition.

They often begin with the same observation:

I do not yet understand why this happens. I would like to find out.