Senior Neuro-Symbolic AI Researcher (Safety)

Huawei Finland R&D; Helsinki; Vor Ort; EUR 90.000 - 130.000; Vollzeit

Benefits dieser Stelle

Occupational healthcare

Culture & sports perks

Phone benefits

Zusammenfassung

Huawei Finland R&D in Helsinki leads the Cloud Content Security Lab, focusing on AI safety, alignment, and scalable detection systems for Harmony OS and related products.

The lab invites a PhD researcher to design novel multimodal AI architectures, improve interpretability, and build robust defenses. You will work with international researchers, publish in top venues, and translate theory into production-grade tools.

Qualifikationen

• PhD in CS, DL, ML, Mathematics or related field.
• Strong focus on AI safety and Explainable AI research.
• Proven proficiency with LLMs/VLMs, neural networks and neuro-symbolic AI.
• Experience with graph reasoning architectures and knowledge graphs.
• Strong programming background in Python, Java or C++ and ML frameworks (PyTorch, TensorFlow).
• Publication record in top conferences (e.g., NeurIPS/ICLR/ICML/AAAI/ACL/ CVPR/ICCV/EMNLP/NAACL).
• Experience in AI alignment to human values and model robustness is valued.
• Ability to work in multinational teams and challenge status quo.

Aufgaben

• Design and implement AI detection architectures for multimodal data (images, videos, text, cultural knowledge).
• Research visual QA, reasoning, and multimodal grounding; explore neural-symbolic methods.
• Develop interpretable, unbiased and aligned AI systems for real-world scenarios.
• Advance alignment techniques for LLMs/CoT/agentic workflows and multi-step reasoning.
• Reverse-engineer neural networks to reveal internal workings and apply explainability tools.
• Create data attribution methods to quantify data influence on predictions.
• Build benchmarks/datasets to measure reasoning quality and explanations.
• Develop defenses against jailbreaking, prompt injections and adversarial exploits.
• Communicate insights to stakeholders and translate research into production-ready code.
• Collaborate to turn AI safety papers into scalable frameworks.

Kenntnisse

LLMs/VLMs

Neural networks

Neural-symbolic AI

Graph reasoning

Python

Java

C++

Explainable AI

Safety

Ausbildung

PhD in Computer Science / Deep Learning / ML / Mathematics

Tools

PyTorch

TensorFlow

Jobbeschreibung

Huawei Finland R&D in Helsinki leads the Cloud Content Security Lab, focusing on AI safety, alignment, and scalable detection systems for Harmony OS and related products.

The lab invites a PhD researcher to design novel multimodal AI architectures, improve interpretability, and build robust defenses. You will work with international researchers, publish in top venues, and translate theory into production-grade tools.