SRLab is a scientific research laboratory advancing responsible, human-centered AI. We build open-source benchmarks, fairness toolkits, and governance frameworks that bridge the gap between cutting-edge AI research and real-world societal impact.
SRLab is a Canadian registered non-profit organization dedicated to building trustworthy, human-centered AI systems that serve society equitably. We develop open-source benchmarks, fairness toolkits, and governance frameworks — ensuring AI technology is fair, transparent, and accountable. Our work bridges the gap between cutting-edge AI research and real-world societal impact, with a commitment to sustainability and public benefit.
Production-ready AI tools, Python packages, datasets, and platforms available for developers and researchers.
The first configurable bias detection and debiasing pipeline for news articles. Detects, recognizes, and replaces biased language using a 3-model Transformer pipeline.
State-of-the-art text analysis and debiasing toolkit. Classifies bias, detects biased entities via NER, and generates debiased text — built on BERT, RoBERTa, and Meta LLaMA-2-7B models.
Open datasets and leaderboards on HuggingFace: SONIC-O1 (4,958 QAs), unbias-plus models, and BEADs — all freely available for researchers and developers.
Span-level social bias classification in text that detects Generalizations, Unfairness, and Stereotypes — going beyond sentence-level detection to pinpoint exact bias spans.
Token-level bias identification framework using a custom BIAS named entity. Works across social media, healthcare, and job hiring domains with transformer-based classification.
A comprehensive fake news detection framework combining traditional ML and state-of-the-art Language Models. Curated dataset of North American election-related news articles.
Clinical application of NLP for detecting COVID-19 risks from text data. Transformer-based approach deployed for real-time public health risk identification and epidemiological monitoring.
Large-scale named entity recognition applied to biomedicine. NLP pipeline for extracting biomedical entities from clinical and research text at population scale.
Comprehensive overview of context-aware recommendation strategies — foundational work covering IoT-integrated systems, temporal dynamics, and user modeling. Published in Computer Science Review (Elsevier).
Enterprise-grade IT infrastructure advisory supporting SRLab's operational resilience. ITIL-based process design, Azure cloud management, cybersecurity compliance, and scalable deployment architecture for research infrastructure at institutional scale.
Peer-reviewed papers in top venues spanning responsible AI, multimodal fairness, AI safety, governance, and network systems.
Open-source tools, benchmarks, and frameworks for building fair, safe, and explainable AI systems.
A real-world benchmark for evaluating multimodal LLMs on audio-video understanding. 231 videos, ~60 hours, 4,958 human-verified QAs across 13 conversational domains with demographic fairness analysis.
A framework for Responsible Reasoning AI Agents — LLM-powered agents that embed fairness, privacy, transparency, and auditability controls throughout every reasoning step, not just the final output.
Tracking the cumulative carbon and water footprint of open-source AI derivatives. Proposes Data and Impact Accounting (DIA) — a coordination mechanism for ecosystem-level environmental visibility.
SRLab (Scientific Research Laboratory for Responsible AI) was founded by Dr. Syed Raza Bashir — a Capstone Supervisor at Sheridan College, Humber College, and Conestoga College, and Research Advisor guiding the lab's strategic direction and institutional partnerships. He brings direct board-level governance and digital transformation experience as former Director of IT (Board-elected) at York Condominium Corporation (YCC 76, 2019–2025), where he led multi-phase cloud migration, IT modernization, and cross-functional governance initiatives — skills directly aligned with large-scale, multi-stakeholder programs like EU Horizon projects. The research program is led by Dr. Shaina Raza — a Top 2% Scientists (Stanford/Elsevier), Responsible AI Leader of the Year, and CIHR Award holder affiliated with the Vector Institute for AI. She serves as Senior Research Advisor, leading the lab's open-source benchmarks, fairness toolkits, and governance frameworks across 100+ peer-reviewed publications. Dr. Khurram Khalid serves as Scientist, and Shujaat Feroze (M.S., University of Wollongong) serves on the Advisory Board for IT Infrastructure — bringing over a decade of enterprise IT leadership across organizations like Telstra, Qantas, Saunders International (ASX:SND), and Comcare (Australian Government). Together with a growing team of Research Interns and graduate students from partner institutions, we collaborate across 6+ institutions globally — publishing in top venues like NeurIPS, EMNLP, ACM, Information Fusion, and Springer.
Recognized globally for leadership in responsible AI research and practice.
Women in AI Summit & Awards 2025 (North America) — for leadership in ethical and trustworthy AI.
Stanford / Elsevier 2024–2025 — recognized among the world's most impactful researchers.
Canadian Institutes of Health Research — supporting responsible AI research in health equity and public health.
Selected as 3 of 137 projects for trustworthy AI — partnering with Vector Institute (Grant 101214389).
Education materials, workshop papers, and posters on AI fairness, sustainability, and agentic safety.
Delivered keynote talks at IEEE SMC, ResearchTrend.AI, and academic/industry venues on Responsible AI and Multimodal Fairness.
Nature reviewer. Editorial member at Springer Discover Computing. NeurIPS main track PC. Guest editor for Elsevier journals.
Co-organized the AI Governance Workshop (2025) — policy, risk, and audit frameworks for trustworthy AI systems.
Active grants, programs, and partnerships aligned with our research in responsible AI, fairness, and governance.
Canada's primary research funding for natural sciences & engineering. Core support for responsible AI research programs.
Government · CanadaPrograms on AI's societal impact, trustworthy AI, and fairness in machine learning systems.
Research Institute · CanadaFollow-on grants from AIXPERT partnership. Trustworthy AI cluster calls for responsible & explainable systems.
International · EUNational Science Foundation programs for fairness, accountability, and transparency. US collaborators can serve as PI.
Government · USAFederal funding for AI safety standards, regulation research, and responsible innovation programs.
Government · CanadaSupports early-career researchers in ML, AI safety, and responsible computing. Up to $60K USD.
Industry · GoogleFunding for open-source, trustworthy AI projects. Great visibility and community alignment.
Foundation · MozillaCollaboration and funding opportunities for responsible AI practices across industry and academia.
Coalition · GlobalSRLab is committed to the highest standards of research integrity and transparency. As a Canadian non-profit, we maintain clear boundaries between our mission and the external institutional affiliations of our team members.
All team members' external affiliations (academic, institutional, or industry) are disclosed publicly on this website. Work conducted at other institutions is clearly attributed to those institutions.
SRLab-funded work and externally-funded work are kept separate. No SRLab resources, data, or funding are used for members' external institutional projects, and vice versa.
Before any grant submission or partnership engagement, potential conflicts of interest are reviewed internally. Team members recuse themselves from decisions where a conflict may arise.
For questions about our governance practices, COI policies, or institutional affiliations, please contact contact@srlab.ai
We're actively seeking grants, institutional partnerships, and industry sponsors to scale our responsible AI benchmarks and governance tools. If you share our vision for trustworthy, human-centered AI — let's talk.