Internship: R&D LLM Mechanistic Interpretability for Privacy - F/M

Mougins Cedex, FR, 06254

SAP

Explore market-leading software and technology from SAP. Become an intelligent, sustainable enterprise with the best in cloud, platform, and sustainability solutions – no matter your industry or size.

View all jobs at SAP

Apply now Apply later

 We help the world run better

At SAP, we enable you to bring out your best. Our company culture is focused on collaboration and a shared passion to help the world run better. How? We focus every day on building the foundation for tomorrow and creating a workplace that embraces differences, values flexibility, and is aligned to our purpose-driven and future-focused work. We offer a highly collaborative, caring team environment with a strong focus on learning and development, recognition for your individual contributions, and a variety of benefit options for you to choose from.

 

What you'll do 
Some recent research activities focus on decomposing LLM activations into sparse linear combinations of features from a large, shared feature dictionary. Using  (unsupervised) sparse autoencoder (SAE) framework allows to find such dictionaries, focusing in particular on monosemantic features (features that activate only for one lement/concept in such dictionary). 
The internship will investigate mechanistic interpretability and design applications to the identification of textual sentences that can leak personal data, such as political options, religious beliefs, sexual preferences etc. In particular, the internship will comprise:
•    Application and replication of recent studies on mechanistic interpretability on real and synthetic datasets.
•    Design of approaches to enhance privacy guarantees of individuals, by detecting sensitive personal data in various contexts (social media contents, business communications etc.).
•    Design of context-dependent approaches to anonymize sensitive personal data, without compromising as much as possible on data integrity and utility.
•    Create clear documentation for evaluated approaches, with the aim of producing a scientific contribution to an international venue.

 

What you bring 
•    University Level: Last year of MSc in Computer Science or beyond
•    Strong Python development skills
•    Natural Language Processing (NLP) Knowledge
•    Machine Learning Knowledge
•    Fluency in English (working language)
•    Abilities in organizing meeting and contacting people
•    Good oral and written communication skills
•    Capacity to write documents in English, ability to synthesize

 

Duration: 6 months
Start date: spring 2025 onwards
In accordance with the French legislation, for any duration over 2 months, this internship is paid in line with the SAP gratification grid.

 

Meet your team 
Customers rely on SAP delivering trustworthy secure products. The “Secure Development Center of Excellence” is responsible for the SAP Secure Software Development and Operations Lifecycle and the respective Product Standard Security and Data Protection.
The team is driving an automation assisted approach in significantly reducing security risks emerging during software development


 

 

​Bring out your best

SAP innovations help more than four hundred thousand customers worldwide work together more efficiently and use business insight more effectively. Originally known for leadership in enterprise resource planning (ERP) software, SAP has evolved to become a market leader in end-to-end business application software and related services for database, analytics, intelligent technologies, and experience management. As a cloud company with two hundred million users and more than one hundred thousand employees worldwide, we are purpose-driven and future-focused, with a highly collaborative team ethic and commitment to personal development. Whether connecting global industries, people, or platforms, we help ensure every challenge gets the solution it deserves. At SAP, you can bring out your best.

We win with inclusion

SAP’s culture of inclusion, focus on health and well-being, and flexible working models help ensure that everyone – regardless of background – feels included and can run at their best. At SAP, we believe we are made stronger by the unique capabilities and qualities that each person brings to our company, and we invest in our employees to inspire confidence and help everyone realize their full potential. We ultimately believe in unleashing all talent and creating a better and more equitable world.
SAP is proud to be an equal opportunity workplace and is an affirmative action employer. We are committed to the values of Equal Employment Opportunity and provide accessibility accommodations to applicants with physical and/or mental disabilities. If you are interested in applying for employment with SAP and are in need of accommodation or special assistance to navigate our website or to complete your application, please send an e-mail with your request to Recruiting Operations Team: Careers@sap.com
For SAP employees: Only permanent roles are eligible for the SAP Employee Referral Program, according to the eligibility rules set in the SAP Referral Policy. Specific conditions may apply for roles in Vocational Training.

EOE AA M/F/Vet/Disability:

Qualified applicants will receive consideration for employment without regard to their age, race, religion, national origin, ethnicity, age, gender (including pregnancy, childbirth, et al), sexual orientation, gender identity or expression, protected veteran status, or disability.
Successful candidates might be required to undergo a background verification with an external vendor.

Requisition ID: 415569  | Work Area: Software-Research  | Expected Travel: 0 - 10%  | Career Status: Student  | Employment Type: Limited Full Time   | Additional Locations: #LI-Hybrid.

Apply now Apply later
Job stats:  3  1  0
Category: Big Data Jobs

Tags: Computer Science LLMs Machine Learning NLP Privacy Python R R&D Research Security

Perks/benefits: Career development Flex hours Team events

Region: Europe
Country: France

More jobs like this