I study how computational systems behave, and build compiler infrastructure and program analysis to make them more reliable and efficient. I like research with a precise question, reproducible experiments and a careful account of what the results do and do not establish.
research interests
- compiler infrastructure & program analysisHow to represent, analyze and transform programs while preserving useful information
- parallel & heterogeneous computingHow to characterize correctness and performance under different configurations
- computational methods for scienceReproducible tools and methods that help understand and improve scientific computing
education
- 2023–present
- B.Sc. in Scientific Computing, Universidad de Medellín, Colombia
publications and talks
- A cognitive architecture for intelligent urban planningAIIA'25 · conference paper · ETH Zürich Research Collection
- Engineering a Hybrid Rust and MLIR Toolchain for AI AgentsEuroLLVM 2026 · Dublin
- Compiling Agentic AI Programs for Dataflow Execution: An MLIR ApproachLLVM Performance Workshop at CGO 2026 · first author, with Rafael Herrera, Isaac Bermúdez and José Monsalve
research and fellowships
- 2026
- LFX Mentorship mentee · RISC-V InternationalArchitectural parameter extraction from RISC-V specifications.
- 2025
- Google Summer of Code contributor · LLVMCompiler artifact analysis and GPU offloading tooling.
- 2025
- Student summer research fellow · ETH ZürichCognitive architectures for urban planning.
professional experience
- present
- Member of Technical Staff · UvicuoFinancial infrastructure for logistics operations.
- 2025–present
- AI Engineer · UrbanEchoComputational systems and urban modeling infrastructure.
- 2024–present
- Security Engineer · Kuruv4Application security assessment and vulnerability analysis.
- 2024–2026
- Data Science Instructor · TripleTenData science and machine learning mentoring.
- 2024
- Security Engineer · Fluid AttacksSecurity tooling and AST-based program analysis.
open source
5 merged contributions across LLVM, Zed, PennyLane, and pandas, plus the LLVM Advisor integration still under review. See the full list.
academic service
- 2024–present
- Co-founder & staff member · LLVM Colombia
- 2024–present
- Artifact evaluation · CGO
- 2026
- Artifact evaluation · MLSys