01 / 05Policy
Federated learning for children's data: encoding trust in education systems
- Venue
- UNICEF Innocenti, Office of Strategy and Evidence
- Year
- 2025
- Role
- Author
- 01
- Researcher
- 02
- AI assurance, data governance
- 03
- Paris, France
- 04
- 48.8566° N, 2.3522° E
Roy Saurabh
Researching the technical and institutional conditions under which consequential AI can be trusted.
Most AI questions begin with “Can the model…?” My work begins later. When that model enters a classroom, a public institution, a sensitive data system, a regulatory environment, what makes it trustworthy?
03 Three worlds, one problem
Each research area is placed by the layers it actually spans.
- AI assurance and accountability
- Privacy-preserving collaborative AI
- Human-centred adaptive systems
- Data governance for public-interest systems
- Machine learning
- Human-computer interaction
- Learning sciences
- Federated learning
- AI accountability
- Children's data
- AI policy
- Evidence
- Public institutions
- Research software
- Audit tooling
- Learning platforms
- Infrastructure
02 / 05Project
An assurance toolchain for deployed AI
- Venue
- Zenodo
- Year
- 2026
- Role
- Author
03 / 05Paper
Data preprocessing and machine learning in wearable data analysis: assessing efficacy and challenges for mental health monitoring of healthcare professionals
- Venue
- Université Paris Cité
- Year
- 2024
- Role
- Author
04 / 05Report
The Transformative Potential of Data for Learning
- Venue
- Broadband Commission for Sustainable Development
- Year
- 2023
- Role
- Writing team
05 / 05Policy
AI in health: how smart technology is breaking new frontiers in medical care
- Venue
- Digital Skills and Jobs Platform, European Commission
- Year
- 2026
- Role
- Author
05 Trajectory
Read as a career, it looks like several. Read as a line of enquiry, it is one.
Institutions
Placed by the year of the first documented relationship. Every line is a real relationship.
Advisory and policy
UNICEF Innocenti2025
Source: UNICEF Innocenti ↗- CurrentExpert advisory group, Good governance of children's data
- 2025Author, Federated learning for children's data
JA Institute2026
Source: JA Institute ↗- CurrentAdvisor, Research Advisory Council
Employment and study
Birla Institute of Technology, Mesra2001
Source: ORCID ↗- 2001-2005B.E., Electronics and Communications Engineering
Tata Consultancy Services2005
Source: ORCID ↗- 2005-2008Systems Engineer
Cedar Consulting / Kyozan2009
Source: ORCID ↗- 2009-2014Director
Zosher · India Skills · Campus Management2014
Source: ORCID ↗- 2014-2018Founder and technology leadership
UNESCO MGIEP2018
Source: UNESCO MGIEP ↗- 2018-2020Chief Technology Officer
- 2019Author, CHI, a knowledge-sharing digital platform powered by machine learning
Centre de Recherches Interdisciplinaires (CRI)2020
Source: ORCID ↗- 2020-2021Lead Data Scientist
Université Paris Cité2020
Source: theses.fr ↗- 2021-2024PhD, Mathematics and Computer Science
- 2020-2021M2, Learning Sciences (EdTech)
- 2024Author, Doctoral research
UNESCO2021
Source: UNESCO ↗- 2021-2025Senior Project Officer
- 2023Writing team, Data for learning
- 2022Writing team, The Transformative Potential of Data for Learning, interim report
AffectLog2025
Publications, talks, records
IEEE2019
Source: Crossref ↗- 2019Co-author, The Alternate Education for the 21st Century
Broadband Commission for Sustainable Development2022
Source: Broadband Commission for Sustainable Development ↗- 2023Writing team, Data for learning
- 2022Writing team, The Transformative Potential of Data for Learning, interim report
European Education and Culture Executive Agency (EACEA)2024
Source: EACEA, European Commission ↗- 2024Panel moderator, AI literacy and critical thinking
CITIC, Universidade da Coruña2025
Source: CITIC, Universidade da Coruña ↗- 2025Invited lecture, Integrating Federated Learning with Data Governance Frameworks for Collaborative and Secure Analysis of Sensitive Data
Digital Skills and Jobs Platform2026
Source: Digital Skills and Jobs Platform (European Commission) ↗- 2026Author, AI in health
Research output
- PaperBA-FedSHAP: A reproducible toolkit for auditing background-induced attribution driftAuthor, SSRN
- PolicyAI in health: how smart technology is breaking new frontiers in medical careAuthor, Digital Skills and Jobs Platform, European Commission
- ProjectAn assurance toolchain for deployed AIAuthor, Zenodo
- SoftwareCSA-Lite, Context-Sliced AI Assurance LiteAuthor, Zenodo
- SoftwareLEXON-Bench, Executable AI Regulatory Obligation ReasoningAuthor, Zenodo
- SoftwareRCEA Passport Engine, Role-Conditioned Evidentiary AdequacyAuthor, Zenodo
- SoftwareProvena, reproducibility package for evidence-laundering stress testsAuthor, Zenodo
- SoftwareBA-FedSHAP, auditing background-induced attribution driftAuthor, Zenodo
- PolicyFederated learning for children's data: encoding trust in education systemsAuthor, UNICEF Innocenti, Office of Strategy and Evidence
- TalkIntegrating Federated Learning with Data Governance Frameworks for Collaborative and Secure Analysis of Sensitive DataInvited lecture, CITIC, Universidade da Coruña, A Coruña, Spain
- PaperData preprocessing and machine learning in wearable data analysis: assessing efficacy and challenges for mental health monitoring of healthcare professionalsAuthor, Université Paris Cité
- TalkAI literacy and critical thinkingPanel moderator, Ninth community workshop on Explainable AI, Brussels, Belgium
Ideas
Field notes
AI governance ·
AI in health: how smart technology is breaking new frontiers in medical care
Where clinical AI is genuinely useful, where it is oversold, and what assurance a regulated care setting has to demand before relying on it.
Published by Digital Skills and Jobs Platform
Data governance ·
Federated learning for children's data: encoding trust in education systems
Published by UNICEF Innocenti
Learning systems ·
CHI, a knowledge-sharing digital platform powered by machine learning
Published by UNESCO MGIEP
Behind every system I study is a person it will beused on, and an institution that has to answer for it.
Let'scomparenotes.
- 01Research collaboration
- 02Policy and advisory work
- 03Invited talks
- 04Academic exchange