Defensive AI Security Intern - Paris, France

Apple

Paris

Hybrid

EUR 20,000 - 27,000

Full time

2 days ago
Be an early applicant

Get a reply from this employer

Your resume and cover letter, tailored to exactly what they’re hiring for.

Get past ATS filters

Job summary

Apple is seeking a Defensive AI Security Intern in Paris to explore inputs that perturb LLM internal states, degrading helpers in attacker analysis of protected code. The role emphasizes mathematical framing of internal-state metrics and gradient-based adversarial techniques, with strict internal documentation and a Paris-based office requirement.

The candidate should be pursuing a Master’s or PhD in CS, math, or related fields, with deep knowledge of transformers, linear algebra, and

Qualifications

  • Strong foundations in linear algebra, optimization (convex and non-convex), and probability/information theory.
  • Solid understanding of transformer architecture internals (attention mechanics, residual stream, embeddings) at a mathematical level.
  • Hands-on experience with gradient-based adversarial attack techniques (FGSM/PGD-style methods, discrete optimization for token sequences, universal adversarial perturbations).

Responsibilities

  • Formulate the problem mathematically: define what "confusion" or degraded internal state means in terms of measurable quantities.
  • Develop gradient-based and optimization-based methods for crafting adversarial inputs targeting internal-state objectives.
  • Analyze transformer internals to identify properties that can be exploited or reinforced.
  • Study transferability of techniques to black-box models and quantify the transfer gap.
  • Build a rigorous evaluation methodology with metrics and ablations.
  • Assess durability under fine-tuning, distillation, or future model versions.
  • Document the framework, methods, and results for Apple's internal use only.

Skills

Linear algebra
Optimization
Probability / Information theory
Transformer internals
Python
PyTorch
Gradient-based adversarial attacks

Education

Masters/PhD in CS or related

Tools

Python
PyTorch

Job description

Defensive AI Security Intern - Paris, France

Paris, Ile-de-France, France Software and Services

Apple's DRM stack protects media content and applications. Across the industry, external attackers increasingly use LLMs — interactively or via automated/agentic pipelines — to accelerate reverse-engineering of software protections; this is a general and growing trend, not specific to any one target. This internship addresses that trend directly: developing, from first principles, inputs that measurably perturb an LLM's internal state (activations, attention distributions, representation geometry) in ways that degrade its ability to usefully assist an attacker analyzing protected code/binaries.

Description

This internship addresses that trend directly: developing, from first principles, inputs that measurably perturb an LLM's internal state (activations, attention distributions, representation geometry) in ways that degrade its ability to usefully assist an attacker analyzing protected code/binaries.Please note: Ideally, we would prefer applications for individuals based within France, specifically Paris. We will require you to access the office in Paris 3 x a week, and 2 days remote working from home. We will not be sponsoring visa's or relocating individuals for this internship.

Responsibilities
  • Formulate the problem mathematically: define what "confusion" or degraded internal state means in terms of measurable quantities (activation/representation drift, attention entropy, logit distribution shift, loss landscape behavior) rather than qualitative output judgments.
  • Develop gradient-based and optimization-based methods for crafting adversarial inputs (e.g. embedding-space perturbation, discrete token optimization à la GCG/greedy coordinate search, universal adversarial triggers) targeting the internal-state objectives defined above.
  • Analyze transformer internals (attention matrices, residual stream, activation subspaces) to identify structural properties — e.g. spectral characteristics, representation collapse, direction vectors linked to specific behaviors — that can be exploited or reinforced.
  • Study whether techniques derived on open-weight models (full white-box access) transfer to closed-weight models via black-box/transfer-based optimization, and quantify the transfer gap mathematically.
  • Build a rigorous evaluation methodology: statistically sound metrics, ablations isolating which mathematical property of the input drives the effect, and robustness checks across model families and versions.
  • Assess durability: whether effects derived from current architectures are expected to hold under fine-tuning, distillation, or future model versions, based on the underlying mathematical mechanism rather than empirical trial-and-error.
  • Document the mathematical framework, methods, and results internally for Apple's use only; no external publication.
Minimum Qualifications
  • Currently completing a Master's degree or enrolled in a PhD, in Computer Science, Applied Mathematics, or a related field, with strong theoretical grounding.
  • Strong foundations in linear algebra, optimization (convex and non-convex), and probability/information theory, and ability to apply them to deep learning internals rather than just use libraries.
  • Solid understanding of transformer architecture internals (attention mechanics, residual stream, embeddings) at a mathematical level.
  • Hands-on experience with gradient-based adversarial attack techniques (FGSM/PGD-style methods, discrete optimization for token sequences, universal adversarial perturbations).
  • Proficiency in Python and PyTorch, including writing custom gradient computations/hooks into model internals (not just using high-level APIs).
Preferred Qualifications
  • Research experience in adversarial machine learning or mechanistic interpretability (activation patching, representation engineering, steering vectors).
  • Familiarity with reverse engineering / binary analysis, to reason about what a real attacker workflow needs from the LLM.
  • Coursework or research in spectral methods, information theory, or statistical learning theory applied to neural network analysis.
  • Prior publications, competition results (e.g. adversarial ML challenges), or open-source contributions demonstrating this kind of mathematical adversarial ML work.

At Apple, we're not all the same. And that's our greatest strength. We draw on the differences in who we are, what we've experienced, and how we think. Because to create products that serve everyone, we believe in including everyone. Therefore, we are committed to treating all applicants fairly and equally. We will work with applicants to make any reasonable accommodations.

At Apple, we believe accessibility is a fundamental human right. You’ll find that idea reflected in everything here — in our culture, our benefits and our digital tools. By welcoming as many perspectives as possible, we help you build a career where you feel like you belong.
Learn about accessibility in Apple’s workplace

Get your free, confidential resume review.

or drag and drop your file here.

Similar jobs

Similar jobs worth comparing

Defensive AI Security Intern — Paris (Hybrid)
Defensive AI Security Intern — Paris (Hybrid)

Apple • Paris

Hybrid
EUR 20,000 - 27,000
Offensive Security Researcher - Userland & Kernel Security
Offensive Security Researcher - Userland & Kernel Security

Apple, Inc. • Paris

Hybrid
EUR 90,000 - 150,000
Offensive Security Researcher - Server Platform & Secure Computing, SEAR
Offensive Security Researcher - Server Platform & Secure Computing, SEAR

Apple, Inc. • Paris

On-site
EUR 150,000 - 210,000
Applied Machine Learning Engineer - Security
Applied Machine Learning Engineer - Security

Apple, Inc. • Paris

On-site
EUR 70,000 - 90,000
AI/ML Engineer - AI Systems for Security, SEAR
AI/ML Engineer - AI Systems for Security, SEAR

Apple, Inc. • Paris

On-site
EUR 90,000 - 150,000
Internship - Machine Learning Research
Internship - Machine Learning Research

Apple, Inc. • Paris

On-site
EUR 13,000 - 20,000
AI/ML Internship: Self-Improving AI Agents for Cybersecurity, SEAR
AI/ML Internship: Self-Improving AI Agents for Cybersecurity, SEAR

Apple, Inc. • Paris

On-site
EUR 10,000 - 17,000
Red Team Platform and Hardware Security Researcher, SEAR
Red Team Platform and Hardware Security Researcher, SEAR

Apple, Inc. • Paris

On-site
EUR 150,000 - 210,000
Offensive Security Researcher, SEAR
Offensive Security Researcher, SEAR

Apple, Inc. • Paris

On-site
EUR 65,000 - 90,000
AIML - Machine Learning Researcher, MLR
AIML - Machine Learning Researcher, MLR

Apple, Inc. • Paris

On-site
EUR 90,000 - 130,000