ManaTEE an OpenSource Private Data Analytics Frame - Common Knowledge Scout

# ManaTEE Open-Source Private Data Analytics Framework > [! note]- > The content of this page is generated by audio/video transcription and text transformation from the content and links of this source. Source: [https://fosdem.org/2025/schedule/event/fosdem-2025-5044-manatee-an-open-source-private-data-analytics-framework-with-confidential-computing/](https://fosdem.org/2025/schedule/event/fosdem-2025-5044-manatee-an-open-source-private-data-analytics-framework-with-confidential-computing/) <video src="https://video.fosdem.org/2025/k4401/fosdem-2025-5044-manatee-an-open-source-private-data-analytics-framework-with-confidential-computing.av1.webm" controls></video> ## Summary & Highlights: ManaTEE is an open-source framework aimed at enabling private data analytics while maintaining data privacy through confidential computing. It integrates Privacy Enhancing Techniques (PETs) to ensure data confidentiality and usability, providing an interactive interface via JupyterLab. The framework supports Trusted Execution Environments (TEEs) to maintain data integrity and execution transparency, facilitating trust between data owners and analysts. ManaTEE simplifies deployment across various confidential computing backends, making secure data analytics scalable and accessible. **Introduction to ManaTEE** ManaTEE addresses the challenges of private data analytics for public research by integrating Privacy Enhancing Techniques (PETs) and confidential computing. It aims to protect data privacy without compromising usability, offering an interactive interface through JupyterLab for researchers. **The Importance of Private Data** Private data is crucial for extracting value in fields like public health, economics, and social sciences. However, privacy risks hinder its use. ManaTEE tackles these challenges, ensuring data confidentiality and fostering trust through TEEs, enabling reproducibility and integrity of research results. **Challenges in Private Data Analytics** ManaTEE identifies key challenges in using private data, including privacy risks, trust issues, and compliance with data protection laws. It emphasizes the need for technical enforcement of privacy policies and accountability in data handling. **The ManaTEE Framework** ManaTEE employs a two-stage execution model, separating data policy and code policy. It uses TEEs to ensure accurate and secure data analytics, enabling flexible data policies during the programming stage and enforcing code policies at the execution stage. **Use Cases and Applications** ManaTEE's framework is applicable in various fields, including public health and civic engagement. It supports cross-organizational data sharing, providing transparency and accountability in research, and is already utilized by platforms like TikTok for public research transparency. ## Importance for an eco-social transformation ManaTEE plays a significant role in eco-social transformation by enabling secure and private data analytics essential for public research, which can lead to societal benefits in public health, economics, and civic engagement. It addresses ethical concerns around data privacy and compliance, providing tools for researchers to access valuable insights without compromising privacy. For eco-social designers, ManaTEE offers a framework to implement privacy-preserving data analytics, fostering trust and collaboration across sectors. Challenges include ensuring compliance with diverse data protection laws and overcoming technical barriers in deploying confidential computing environments. ## Slides: | | | | --- | --- | | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_001.jpg\|300]] | ManaTEE is introduced as an open-source framework designed to facilitate private data analytics with a focus on confidential computing. Presented by Dayeol Lee from TikTok, the framework aims to enable secure data handling in public research, addressing privacy concerns while maintaining usability. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_002.jpg\|300]] | The presentation outlines the contents, starting with the challenges of private data analytics and the need for robust solutions. It covers existing approaches, introduces the ManaTEE project, and includes a tutorial demo, discussing both current and future project directions. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_003.jpg\|300]] | This section discusses the concept of private data analytics, emphasizing its importance and the challenges associated with maintaining data privacy while enabling valuable insights for public research. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_004.jpg\|300]] | Private data is highlighted as crucial for value extraction and public interest, particularly in fields like public health and civic engagement. The slide discusses the dual importance of private data for both business and societal benefits. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_005.jpg\|300]] | Reiterating the importance of private data, the slide emphasizes its role in extracting value and serving public interest, highlighting its potential impact on various research domains. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_006.jpg\|300]] | The slide emphasizes the significance of sharing private data for public interest, detailing its applications in public health, safety, education, and civic engagement. It underscores the need for secure data sharing practices to enable beneficial insights. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_007.jpg\|300]] | Cross-organizational data sharing is often required for understanding complex issues, such as illicit drug promotion. The slide provides an example of research leveraging cross-platform data to gain insights into such activities. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_008.jpg\|300]] | The slide introduces HDR UK as a trusted research environment aimed at facilitating secure access to health data for public research, illustrating efforts to build trusted frameworks for data sharing. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_009.jpg\|300]] | This section outlines the difficulties in private data analytics, focusing on the inherent challenges of data privacy, trust, and compliance with legal and organizational policies. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_010.jpg\|300]] | Data privacy risks are highlighted as a major challenge, with issues such as trust conflicts, data abuse, and compliance with privacy policies. Legal restrictions on data handling and location changes are also discussed. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_011.jpg\|300]] | The slide discusses the challenges of accountability and transparency in data analytics, emphasizing the need for clear responsibility and verifiability in data handling across distributed systems. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_012.jpg\|300]] | This section outlines the requirements for a standard approach to private data analytics, emphasizing the need for strong privacy protection, technical policy enforcement, accountability, and usability. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_013.jpg\|300]] | A 'standard' approach to private data analytics is needed, providing strong privacy protection, technical policy enforcement, accountability, and usability. The slide emphasizes proactive measures and system auditability. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_014.jpg\|300]] | The section introduces existing approaches to private data analytics, setting the stage for discussing the ManaTEE framework and its advantages over current solutions. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_015.jpg\|300]] | Existing industry solutions for security and privacy are discussed, including SQL-based data clean rooms, differential privacy, and trusted execution environments. Each approach's strengths and limitations are highlighted. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_016.jpg\|300]] | The slide evaluates the technical difficulties of existing solutions, comparing differential privacy, SQL-based data clean rooms, and trusted execution environments in terms of enforcement, transparency, usability, and accuracy. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_017.jpg\|300]] | | | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_018.jpg\|300]] | ManaTEE's goals include enforcing privacy policies via PETs, providing usability through interactive tools, ensuring accuracy and transparency, and simplifying deployment. The framework aims to make private data analytics accessible and reliable. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_019.jpg\|300]] | Different stages of data analytics require distinct approaches. The programming stage involves smaller data sets and higher privacy risks, while the execution stage handles larger data and is easier to control, guiding ManaTEE's two-stage approach. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_020.jpg\|300]] | ManaTEE employs a two-stage data clean room approach, using synthetic data during programming and actual data during execution. This model separates data and code policies, enabling secure and accurate analytics. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_021.jpg\|300]] | The benefits of ManaTEE's two-stage execution model include flexible data policies during programming and secure, accurate results during execution. Confidential computing ensures trust and execution integrity, enabling cross-organizational collaboration. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_022.jpg\|300]] | The slide outlines ManaTEE's data and code pipeline, detailing the use of JupyterLab, data SDK, and executor pods in a TEE backend. It highlights the framework's flexibility and ease of deployment. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_023.jpg\|300]] | ManaTEE supports easy cloud deployment via Terraform, leveraging resources like containers, key management, and cloud storage. The framework's infrastructure is designed for scalability and efficiency. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_024.jpg\|300]] | This section introduces potential use cases for ManaTEE, demonstrating its applicability in various research fields and industries, enhancing transparency and accountability in data analytics. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_025.jpg\|300]] | ManaTEE provides transparency to researchers in fields like public health, economics, and civic engagement, enabling insights through secure and private data analytics. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_026.jpg\|300]] | TikTok's research tools, including the Virtual Compute Environment (VCE), are highlighted as applications of ManaTEE, showcasing the framework's role in providing transparency and facilitating public research. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_027.jpg\|300]] | Potential use cases for ManaTEE include ads and marketing, machine learning, and AI model evaluation, demonstrating its versatility in handling private data across various domains. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_028.jpg\|300]] | The tutorial demo provides a practical introduction to ManaTEE, guiding users through the framework's setup and usage for private data analytics. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_029.jpg\|300]] | The demo scenario involves training a model to predict insurance charges using a Kaggle dataset. Differentially-private synthetic data is used in the first stage to protect privacy, showcasing ManaTEE's application. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_030.jpg\|300]] | This slide provides a visual overview of ManaTEE, summarizing its components and workflow in the context of private data analytics. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_031.jpg\|300]] | ManaTEE's current and future directions are outlined, highlighting its evolution from single-source data collaboration to multi-way, cross-organizational collaboration, and advancements in backend and compute capabilities. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_032.jpg\|300]] | The project timeline details ManaTEE's development milestones, including its launch, open sourcing, renaming, and community engagement, illustrating its growth and collaborative efforts. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_033.jpg\|300]] | ManaTEE's first community release is announced, offering resources like tutorials, local deployment guides, and release notes, inviting collaboration and experimentation. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_034.jpg\|300]] | The slide invites collaboration on the ManaTEE project, providing links to Google Groups and GitHub for community engagement and contribution. | ![[FOSDEM 2025/assets/ManaTEE-an-OpenSource-Private-Data-Analytics-Frame/preview_035.jpg\|300]] | The Q&A section invites further discussion, providing contact information for continued engagement and exploration of ManaTEE's capabilities. ## Links [ManaTEE slides in pdf](https://fosdem.org/2025/events/attachments/fosdem-2025-5044-manatee-an-open-source-private-data-analytics-framework-with-confidential-computing/slides/237870/FOSDEM_20_qJOxKd0.pdf) [Video recording (AV1/WebM)](https://video.fosdem.org/2025/k4401/fosdem-2025-5044-manatee-an-open-source-private-data-analytics-framework-with-confidential-computing.av1.webm) [Video recording (MP4)](https://video.fosdem.org/2025/k4401/fosdem-2025-5044-manatee-an-open-source-private-data-analytics-framework-with-confidential-computing.av1.mp4) [Video recording subtitle file (VTT)](https://video.fosdem.org/2025/k4401/fosdem-2025-5044-manatee-an-open-source-private-data-analytics-framework-with-confidential-computing.av1.vtt)