amatria.dev
Iñaki
Amatria Barral
Software engineer with a focus on compilers, LLVM, and HPC. I enjoy hard problems.
About
Software engineer with a focus on compilers, LLVM, and HPC. I work mostly in modern C++, extending frontends, writing analysis passes, and optimizing code at a low level. I have contributed to projects like Clang, Flang, and tree-sitter, and I enjoy the kind of problems where performance actually matters, from HPC tooling used at national research labs to genome-scale bioinformatics.
- Languages: C++, C, Fortran, Python, Lua.
- Compiler stack: LLVM, Clang, Flang, Bear, tree-sitter.
Experience
Compiler Engineer
Codee
- Developed LLVM IR analysis and transformation passes for performance diagnostics and auto-remediation, enabling adoption by national research labs including Argonne, Oak Ridge, and NERSC.
- Extended Clang and Flang frontends for AST analysis and semantic diagnostics on real-world HPC codebases (Fortran, C, and C++), including targeted MLIR modifications to support compiler extensions.
- Actively contributed upstream to LLVM, Bear, and tree-sitter-fortran; patches merged into production and issues reported and resolved across the broader compiler community.
Research Assistant
Universidade da Coruña
- Achieved 200x speedup on lncRNA-DNA interaction prediction, enabling transcriptome-scale analysis of the human genome in under one hour (Bioinformatics, D1).
- Reduced a 3-month computation to under 20 hours on a modest cluster via a key memory optimization in the parallelization of RIblast (Future Generation Computer Systems, D1; Methods in Molecular Biology, D1).
- Developed data processing pipelines and statistical analysis workflows in Python to validate experimental results across large genomic datasets.
- All work released as open source; tools actively maintained and used by researchers worldwide.
Research Support Assistant
Universidade da Coruña
- Conducted research in bioinformatics and HPC.
- Implemented and evaluated algorithms for scientific workloads.
Undergraduate Software Engineer
Nomasystems SL
- Developed and deployed a QA pipeline for an iOS application used at scale.
- Automated testing and validation processes.
Open Source
- Designed and implemented a compiler from scratch for a Lua-like language.
- Implemented LLVM IR backend for native code generation.
- Focused on modular design, clarity, and performance (C++17).
- Increased grammar robustness for real-world codebases.
- Implemented and reviewed patches in the Flang frontend, including semantic analysis, module handling, and diagnostics.
- Fixed non-deterministic behavior in Fortran module generation and aligned driver behavior with gfortran via the -x flag.
- 5+ merged patches and 10+ resolved issues.
- Genome-wide tool for lncRNA-DNA triplex helix detection published in Bioinformatics (D1) with a demonstrated 200x speedup.
Publications
Methods in Molecular Biology
Iñaki Amatria-Barral et al.
38th ACM/SIGAPP Symposium on Applied Computing
Iñaki Amatria-Barral et al.
Bioinformatics
Iñaki Amatria-Barral et al.
Future Generation Computer Systems
Iñaki Amatria-Barral et al.
Education & Training
International HPC Summer School
Georgia Institute of Technology
Advanced program · Scalable computing, parallel programming, performance optimization.
M.S. in High-Performance Computing
Universidade da Coruña
Graduated with honors · Academic excellence award · Vectorization, memory optimization, parallel programming, GPU computing.
B.S. in Computer Science
Universidade da Coruña
Graduated with honors · Academic excellence award.