Junior Algorithm Researcher

Huawei Theory Lab is looking for a top student talent to join the Complex Systems and Control group in Warsaw. This position invites a motivated MSc or BSc candidate to participate in the development of hardware-accelerated machine-learning-assisted optimization methods. The project sits at the intersection of optimization algorithms, physics-based modelling, and applied machine learning, and targets a methods paper in a high-impact journal.

• Currently enrolled in an MSc or BSc program in Physics, Applied Mathematics, Computer Science, Statistics, Engineering, or related fields.
• Good knowledge (coursework or project experience) in at least one of the following: optimization (combinatorial optimization, QUBO, integer programming, linear programming), numerical methods for differential equations or statistical physics, or machine learning (graph neural networks, time-series prediction, deep generative models).
• Strong programming skills in Python and at least one of PyTorch or NumPy; ability to navigate unfamiliar codebases.
• Comfortable with Git, terminal, and reproducible research practices.
• Excellent teamwork spirit and communication skills.

Nice to have:

• Background in protein design, structural bioinformatics, or molecular modelling.
• Experience running PyTorch on accelerators (GPU, NPU).
• Experience with diffusion models training and inference solvers (DDPM, DDIM, DPM-Solver) and familiarity with diffusion theory.

• Conduct applied research in high-performance optimization, focusing on novel Ising/QUBO formulations and hardware-accelerated solvers (simulated annealing, simulated bifurcation, coherent Ising machines) on NPU and GPU platforms.
• Mathematically model real-life optimization scenarios, translating protein backbone geometries and energy functions into constrained QUBO/Ising cost landscapes and benchmarking against deep-learning baselines on public benchmarks.
• Design, deploy, and test optimization algorithms; characterize solver performance across varied problem instances, hardware backends, and parameter regimes.
• Run systematic benchmarks to understand solver behavior, analyzing interactions between algorithm parameters, problem size, and hardware, and apply control-theory-inspired parameter scheduling to improve convergence.
• Reproduce and extend results from relevant literature, occasionally proposing new variants or optimization directions.
• Actively contribute to the project work plan, collaborating closely with other team members and groups within Huawei.