Freelancer profile translated to English.
Back to original languageAbout Abdelhafid
President and Principal Architect at PerfScale Systems, I support startups, research labs, and AI/HPC players in hardware/software co-design and extreme performance optimization (Workload Acceleration).
My expertise lies at the intersection of silicon and code: I work from micro-architectural signature detection to global performance optimization of compute clusters (CPU/GPU).
Areas of intervention:
• Micro-architectural performance audits & advanced telemetry.
• Kernel optimization (CUDA, OpenMP) and LLM infrastructures (llama.cpp, PyTorch).
• Elimination of system noise through the design of deterministic benchmarking frameworks.
• Diagnosis of complex bottlenecks and TCO optimization (perf/watt).
Collaboration modalities:
Expert interventions, short audits, or long R&D support, billed directly through my B2B structure (PerfScale Systems).
English
Native or bilingual
French
Native or bilingual
Can work on-site
Paris (up to 50km)
Experience
- PerfScale SystemsSenior HPC / AI Systems Performance Architect — Co-design & Low-Level Optimization ExpertTECHMay 2026 - Today (2 months)Sèvres, France
- Performed advanced system-wide performance audits, end-to-end profiling, and workload optimization to maximize infrastructure efficiency and reduce TCO.
- Expertise in hardware-software co-design for interfacing low-level compute units with software layers, developing drivers/APIs, and accelerating hardware interaction loops.
- Designed and deployed custom deterministic benchmarking frameworks to eliminate system noise, isolate micro-architectural signatures, and ensure reliable performance baselines.
- IntelSenior Staff Engineer — Performance Tools ArchitectTECHSeptember 2022 - April 2026 (3 years and 7 months)Boulogne-Billancourt, France
- Led the design of an automated multi-execution framework for Computational Quality Analysis (QaaS) & performance optimization, achieving >3X gains on compilers, runtime scalability, hardware architectures, and data volumes.
- In-depth micro-architectural characterization of large-scale workloads and inference engines (llama.cpp) on x86 and ARM CPU architectures.
- Isolated critical frequency throttling issues (AVX-512) using advanced system telemetry.
- Conducted rigorous comparative analyses of energy efficiency (perf/watt) for Deep Learning workloads to optimize datacenter TCO.
- Bull / AtosR&D Software Engineer — Technical LeadTECHJanuary 2016 - January 2022 (6 years)Les Clayes-sous-Bois, France
- Designed and developed scalable runtime systems for advanced power management on TOP500 Supercomputers.
- Optimized large-scale production HPC applications (NEMO, Gromacs, HPCG) through dynamic hardware reconfiguration (DVFS & Power Capping), enabling up to 20% energy savings.
- Developed in-situ characterization methods using Linux perf and PMPI to capture real-time computation, memory, and communication phases of applications.
- Defined R&D scientific roadmaps, managed deep-tech European collaborative projects, and integrated automated performance regression tools into CI/CD production pipelines.
Recommendations
Be the first to recommend Abdelhafid
Help this freelancer shine by sharing your experience working together.
These freelancer profiles also match your criteria
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Education
- Doctorate (PhD)University of Versailles2012Doctorat (PhD)
- Master in Computer ScienceUniversity of Versailles2008Master en Informatique