Requisition ID: 273917
Work Area: Information Technology
Expected Travel: 0 - 10%
Career Status: Student
Employment Type: Limited Full Time
SAP started in 1972 as a team of five colleagues with a desire to do something new. Together, they changed enterprise software and reinvented how business was done. Today, as a market leader in enterprise application software, we remain true to our roots. That’s why we engineer solutions to fuel innovation, foster equality and spread opportunity for our employees and customers across borders and cultures.
SAP values the entrepreneurial spirit, fostering creativity and building lasting relationships with our employees. We know that a diverse and inclusive workforce keeps us competitive and provides opportunities for all. We believe that together we can transform industries, grow economics, lift up societies and sustain our environment. Because it’s the best-run businesses that make the world run better and improve people’s lives.
ABOUT US (TEAM)
Maintaining security is a constantly shifting task, and we need to respond with continuous learning and research. The portfolio of SAP Security Research contains those topics that we believe are most important for SAP’s security future.
SAP’s vision to secure business is built on 3 ideals: Zero-Vulnerability, to harden the software by eliminating vulnerabilities, Defensible Application, to enable the software to identify and prevent attacks, and Zero-Knowledge, to make any theft of data useless through encryption.
Considering these aspects, SAP Security Research covers the following focal areas: Anonymization for Big Data, Secure Internet of Things, Software security analysis, Open-source analysis, Deceptive application, Applied cryptography, Quantum technology, and Machine Learning as enabler for the next generation of security.
PURPOSE AND OBJECTIVES
This internship is based in the SAP Labs France Research Lab, in Sophia-Antipolis. The work will be performed in the context of the Research Program “Security & Trust”, under the Threat Intelligence topic.
With the recent developments in the information technologies, companies have adopted deep learning techniques to transform their manual tasks into automated services in order to increase the service quality for their customers (Ge, Zhiqiang et. al). Although this transformation brings some new business opportunities for technology companies, it comes with the burden of competing in a rapidly changing business arena. In order to make profit from these business opportunities, these companies need to be in good positions in the marketplace, which requires vast amount of resource allocation to train a deep learning model such as money, time, hardware, expert knowledge, etc. Thus, since productionized models becomes intellectual properties for these technology companies, their protection is a challenging issue when they are deployed on publicly accessible platforms (Zhang, Jialong et al.). Watermarking techniques are one of the most suitable candidates to overcome this issue. Similar to image watermarking, markers are embedded into the deep learning model during the training phase and the owner of the model can easily verify its model by feeding these markers with a sequence of inference queries.
The goal of the internship is to develop a platform for sharing machine learning models (for inference or research purposes) while implementing watermarking techniques to preserve the ownership of the models. The watermarking process should be compatible with all types of models / sources of data and should be completely transparent for the end-user.
Ge, Zhiqiang and Song, Zhihuan and Ding, Steven X and Huang, Biao (2017), Data mining and analytics in the process industry: The role of machine learning.
Zhang, Jialong and Gu, Zhongshu and Jang, Jiyong and Wu, Hui and Stoecklin, Marc Ph. and Huang, Heqing and Molloy, Ian (ASIACCS 18’), Protecting Intellectual Property of Deep Neural Networks with Watermarking
EXPECTATIONS AND TASKS
In This Internship, The Student Will
Study State-of-the-Art on Watermarking deep neural networks.
Propose a solution to develop a ML platform for sharing and protecting ML models.
Implementation of a Proof-of-Concept demonstrating the feasibility of the solution.
We expect that 60% of time will be dedicated to development and 40% to research activities.
PROFILE/EDUCATION/SKILLS AND COMPETENCIES
University Level: Last year of MSc in Computer Science or beyond
Exposure to industry or academic research, particularly in deep learning, neural networks, or related fields.
Experience with one or more deep learning libraries and platforms (e.g., TensorFlow, Caffe, Chainer or PyTorch).
Experience building distributed ML pipelines frameworks.
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.
WHAT YOU GET FROM US
Success is what you make it. At SAP, we help you make it your own. A career at SAP can open many doors for you. If you’re searching for a company that’s dedicated to your ideas and individual growth, recognizes you for your unique contributions, fills you with a strong sense of purpose, and provides a fun, flexible and inclusive work environment – apply now.
SAP'S DIVERSITY COMMITMENT
To harness the power of innovation, SAP invests in the development of its diverse employees. We aspire to leverage the qualities and appreciate the unique competencies that each person brings to the company.
SAP is committed to the principles of Equal Employment Opportunity and to providing reasonable accommodations to applicants with physical and/or mental disabilities. If you 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 (Americas: Careers.NorthAmerica@sap.com or Careers.LatinAmerica@sap.com, APJ: Careers.APJ@sap.com, EMEA: Careers@sap.com).
Successful candidates might be required to undergo a background verification with an external vendor.