Shaina Raza, PhD
Responsible AI Leader

Applied Machine Learning Scientist at Vector Institute for Artificial Intelligence, Toronto · Expert in LLMs, Multimodal AI, Fairness & Bias, Green AI, and AI Governance

🏆 Responsible AI Leader of the Year 2025 Top 2% Scientist — Stanford/Elsevier 2025 & 2026 EU Horizon AIXPERT Partner (€515K) Highly Cited — Thousands of Citations ICML 2026 Spotlight
Dr. Shaina Raza

Shaina Raza, PhD

Applied ML Scientist · Responsible AI
Vector Institute, Toronto

3K+
Citations
65+
Publications
30+
Keynotes
15+
Mentored

News & Updates

Nov 2026
Panelist — Accelerate Her Future Virtual Summit 2026Panel
"Who benefits from AI? Who's left out?" Nov 20, with Dr. Tanya Chung-Tiam-Fook, moderated by Helena Yu.
Nov 2026
Speaker — MLOps North Toronto Summit (TMLS)Talk
"Responsible AI Beyond Accuracy: Fairness, Safety, and Sustainability", Evals & Benchmarks track, Nov 6. Event →
Oct 2026
Speaker — MCP Dev Summit TorontoTalk
Presented Vector's work on agentic AI evaluation, harness engineering, and responsible AI.
Oct 2026
Recognized among Top 2% Scientists Worldwide for the second consecutive year — Stanford / Elsevier 2026 Update.Top 2% Database →
Oct 2026
Deepfake videos and municipal elections: what voters should watch forMedia
Interviewed by Metroland (Loraine Centeno) on spotting deepfake election videos; syndicated across Ontario (InsideHalton, Niagara Falls Review, Waterloo Region Record, OurWindsor). Article →
Oct 2026
EMNLP 2026 — 2 acceptancesEMNLP
· SONIC-O1 — Main Conference arXiv →
· DiagFlowBench — Industry Track, Budapest arXiv →
Oct 2026
NeurIPS 2026 — 3 workshop papers accepted (Sydney, Dec 6–12)NeurIPS
· Stress-Testing Efficient Responsible-AI Evaluation: When Compute Savings Change Benchmark Conclusions — TAE Workshop arXiv → Project →
· Detecting Deception, Not Deepfakes: Why Media Forensics Needs Social Theories — TAE Workshop arXiv →
· AgentFinVQA: An Auditable, On-Premise Multi-Agent Pipeline for Financial Chart QA — AABA4ET Workshop Project →
Oct 2026
Journal and conference acceptancesPublished
· HumaniBench — ACM Transactions on Intelligent Systems and Technology DOI →
· Evaluating and Regulating Agentic AI — Information Fusion (2026) Project →
· Detecting and Reasoning About Bias in Multimodal Content — IASEAI 2026 Paper →
Sep 2026
New preprints on evaluating trustworthy and agentic AIPreprints
· A Unified Evaluation Framework for Trustworthy LLMs, Agentic AI, and Multimodal Systems arXiv →
· Harness-Aware Evaluation of LLM Agents Preprint →
· FairLens: Fairness in Vision-Language Models arXiv →
· Transparency in Agentic AI survey Project →
Aug–Sep 2026
Featured in Nosian Magazine and SmarterArticlesMedia
Quoted on the hidden cost of AI in "AI's Unasked Question: What Did That Cost?" →; deepfake research cited in "Your Face Is Not Yours" →
Aug 2026
Deepfakes in the Foundation-Model Era: A Survey of Forensics, Generation and Distribution Across the Social Media LifecycleSurvey
Survey of generative audio-video deepfakes across the full social media lifecycle: creation, distribution, detection, provenance, and remediation. Preprint →
Jul 2026
AIXPERT feature: "Can AI see the world without stereotypes?"Feature
Consortium feature on our multimodal fairness research; also profiled in the AIXPERT Summer newsletter and "Inside AIXPERT: One Year of Progress". Feature →
Jun 2026
UnBias-Plus released as a free, open-source toolkitRelease
Covered by BetaKit, BNN Bloomberg, SecurityBrief, Tech Talent Canada, Canadian Healthcare Technology, BornCity, and AI Loop. BetaKit → Paper →
2026
Vector blog: "Mixture of Experts: From Sparse Routing to Multimodal Deployment"Blog Read →
Jun 2026
Vector blog: "When AI watches and listens: Introducing SONIC-O1"Blog Read →
Jun 2026
Generative Deepfake Videos in the Foundation-Model Era: A Timeline of Eroding Trust in Visual EvidenceACM MAD '26
Published at the 5th ACM International Workshop on Multimedia AI against Disinformation (MAD '26), Amsterdam. DOI →
Jun 2026
UnBias-Plus showcased to the Prime Minister of Canada, Mark CarneyShowcase
Our next-generation bias detection & debiasing toolkit (Qwen3-4B) was presented to Prime Minister Mark Carney. Project →
Jun 2026
Vector Institute–Helmholtz Munich MOU signedPartnership
New German–Canadian partnership for joint AI & ML research, researcher exchanges, and bilateral funding. Presented at HAICON 2026 (Helmholtz Munich, Jun 2026) in the UNLOCK benchmarking session. News →
Jun 2026
Named to The Peak's Emerging Leaders 2026 (Artificial Intelligence) — recognizing the up-and-coming leaders shaping Canada's economy, culture, and society.Recognition The Peak →
May 2026
The Deepfakes We Missed: We Built Detectors for a Threat That Didn't ArrivePosition
Position paper arguing that deepfake detection research still targets a late-2010s threat model while intimate-image abuse, voice-clone scams, and emotional fraud dominate real-world harm. arXiv →
May 2026
Keynote Speaker — GREEN-AI Workshop @ ICML 2026 (Helmholtz AI)ICML 2026 Spotlight
Presenting "Sustainable Open-Source AI Requires Tracking the Cumulative Footprint of Derivatives" — introducing the Data and Impact Accounting (DIA) framework. Accepted as a Spotlight Position Paper at ICML 2026. Paper →
Feb 2026
From Features to Actions: Explainability in Traditional and Agentic AI SystemsAgentic Eval
Trace-grounded, trajectory-level evaluation that localizes where agentic AI systems break down — where attribution methods cannot. arXiv →
Jan 2026
SONIC-O1 released — a fully human-verified benchmark evaluating multimodal LLMs on real-world audio-video understanding across 13 domains (summarization, evidence-grounded MCQ reasoning, temporal localization), with demographic/fairness analysis.Benchmark Project →
Jan 2026
F-DPO: Reducing Hallucinations in LLMs via Factuality-Aware Preference LearningFindings 2026
Factuality-aware Direct Preference Optimization — cuts hallucination rates up to 5× (Qwen3-8B) with no auxiliary reward model. Accepted to Findings (2026). arXiv →
Jan 2026
Panelist — AAAI 2026 AI Governance Workshop (AIGOV), SingaporeTalk
AI Governance, Safety and Responsible AI panel alongside IBM Research, Google DeepMind, Microsoft Research, Kush Varshney, and Francesca Rossi. Workshop →
Jan 2026
Journal: VLDBench: Evaluating Multimodal Disinformation with Regulatory AlignmentIF 18.6
Information Fusion (2026) — Q1, IF 18.6. Project →
Jan 2026
Journal: TRiSM for Agentic AI: Trust, Risk and Security in LLM-based Multi-Agent Systems
AI Open (2026). DOI →
Jan 2026
Journal: Review-based Recommender Systems: Survey of Approaches, Challenges and Future Perspectives
ACM Computing Surveys (2026). DOI →
Oct 2025
NeurIPS 2025 — 6 AcceptancesNeurIPS
Education ×2, Workshops ×1, Posters ×3:
· From Data to CO₂: The Hidden Climate Cost of AI — Education Materials ↗
· Prompting Away Stereotypes? Evaluating Bias in Text-to-Image Models — GenProCC Workshop
· Trust, Risk and Security in Agentic AI — LAW@NeurIPS
· Bias in the Picture: Benchmarking VLMs with Social-Cue News Images — Poster
· LinguaMark: Do Multimodal Models Speak Fairly? — Poster
· HumaniBench: Human-Centric Framework for Multimodal Models — Poster
Oct 2025
Co-organizer, AI Governance Workshop — policy, risk, and audit for trustworthy AI. Site →
Oct 2025
Keynote at IEEE SMC 2025 — Responsible AI and Multimodal FairnessKeynote
Oct 2025
Recognized among Top 2% Scientists Worldwide — Stanford / Elsevier 2025 Update.Top 2% Database →
Sep 2025
Appointed Editorial Member, Springer Discover Computing Journal. Journal →
Jul 2025
Jun 2025
🏆 Winner — Responsible AI Leader of the Year, Women in AI Summit & Awards 2025 (North America)Award
Jun 2025
AIXPERT (EU Horizon) launched — Vector Institute selected as 1 of 3 from 137 global applicants. Grant No. 101214389, €515K. CORDIS →
Apr 2025
Journal: Developing Safe and Responsible LLMsIF 7.5
Machine Learning, Springer (2025). Paper →
Mar 2025
Green AI Benchmark released: Benchmarking Energy Footprints of AI Models

Research Projects

🌐

AIXPERT — EU Horizon

Trustworthy AI for industry. Vector Institute selected as 1 of 3 from 137 global applicants. Grant No. 101214389, €515K.

aixpert-project.eu →
⚖️

FairSense-AI

Comprehensive toolkit for fairness, sustainability, and explainability in large-scale AI systems. Bias auditing, emissions reduction, responsible AI compliance.

FairSense-AI →
🔬

HumaniBench

Human-centric benchmark evaluating fairness, social bias, and empathy in large multimodal models. Multilingual and adversarial test cases. Published in ACM TIST (2026).

ACM TIST →
🧪

F-DPO

Factuality-aware Direct Preference Optimization — reducing hallucinations in LLMs (up to 5× on Qwen3-8B) with no auxiliary reward model. Accepted at ACL 2026 Findings.

arXiv:2601.03027 →
📊

VLDBench

Benchmark for evaluating vision-language models on multimodal disinformation detection with regulatory alignment. Published in Information Fusion (IF 18.6).

VLDBench →
🎬

SONIC-O1

Fully human-verified benchmark evaluating multimodal LLMs on real-world audio-video understanding — summarization, evidence-grounded MCQ reasoning, and temporal localization across 13 domains, with demographic/fairness analysis. Accepted at EMNLP 2026 (Main).

sonic-o1 →
♻️

Green AI & Sustainability

Energy footprint benchmarking of AI models, sustainable LLM training, and carbon reduction research. Spotlight at ICML 2026 (Helmholtz AI Workshop).

GreenBench →
🛡️

UnBias-Plus

Next-generation bias detection and debiasing toolkit powered by fine-tuned LLMs (Qwen3-4B). Detects biased language segments, classifies type and severity, suggests neutral rewrites, and returns a fully rewritten unbiased article — all in one structured output.

GitHub →
🤖

FairSense-AgentiX

Agentic fairness and AI-risk analysis platform using ReAct reasoning loops with dynamic tool selection and self-critique. Multi-modal bias detection across text, images, and datasets. Production-ready FastAPI + WebSocket streaming + React UI. Funded by EU Horizon AIXPERT. JOSS paper under review.

GitHub →
💉

Model Immunization

Techniques for immunizing LLMs against falsehoods and harmful content, improving robustness and factuality. Accepted at IEEE WCCI/IJCNN 2026.

arXiv:2505.17870 →
🧭

Agentic Evaluation

Evaluating and Regulating Agentic AI (Information Fusion 2026), From Features to Actions (trajectory-level explainability), Harness-Aware Evaluation, a Unified Evaluation Framework for trustworthy LLMs and agents, and the Transparency in Agentic AI survey.

arXiv:2602.06841 →
🎭

Deepfakes in the Foundation-Model Era

Research program on generative audio-video deepfakes: position paper on misallocated detection effort, ACM MAD '26 timeline paper, lifecycle survey of forensics and distribution, and deception-centred evaluation (NeurIPS 2026 TAE Workshop).

Survey →
⚡

Sustainable Responsible-AI Evaluation

Does cheaper evaluation (batching, INT8/INT4 quantization, benchmark subsetting) preserve accuracy, bias, and fairness conclusions on BBQ and BBQ-V, and what does it cost in energy, CO₂e, and water. NeurIPS 2026 TAE Workshop.

Project →
🔍

FairLens

Benchmark for fairness in vision-language models across hiring, legal, and healthcare decision contexts.

arXiv:2609.01691 →
🏥

Equity in Public Health AI

Embedding fairness and equitable decision-making in AI systems for public health surveillance. CIHR HSI Fellowship ($175K) + AI4PH Award ($20K/yr).

CIHR Fellowship →

Selected Publications

Showing recent high-impact work. View all on Google Scholar →

SONIC-O1: A Real-World Benchmark for Evaluating Multimodal LLMs on Audio-Video Understanding
Radwan, A. Y., Emmanouilidis, C., Tabassum, H., Pandya, D., Raza, S. · EMNLP 2026 (Main) · arXiv →
DiagFlowBench: Evaluating LLMs on Flowchart-Grounded Diagnostic Dialogues
Gil de Avalle, G., Emmanouilidis, C., Maruster, L., Raza, S. et al. · EMNLP 2026 (Industry Track) · arXiv →
HumaniBench: A Human-Centric Framework for Large Multimodal Models Evaluation
Raza, S. et al. · ACM Transactions on Intelligent Systems and Technology (2026) · DOI →
Evaluating and Regulating Agentic AI
Raza, S. et al. · Information Fusion (2026)IF 18.6 · Q1 · Project →
Detecting and Reasoning About Bias in Multimodal Content
Raza, S. et al. · IASEAI 2026 · Paper →
Stress-Testing Efficient Responsible-AI Evaluation: When Compute Savings Change Benchmark Conclusions
El Kady, A., Narayanan, A., Riaz, R., Ioannou, Y., Raza, S. · NeurIPS 2026, TAE Workshop · arXiv →
Detecting Deception, Not Deepfakes: Why Media Forensics Needs Social Theories
Ho, J. et al. (incl. Raza, S.) · NeurIPS 2026, TAE Workshop · arXiv →
AgentFinVQA: An Auditable, On-Premise Multi-Agent Pipeline for Financial Chart QA
Narayanan, A., Raza, S. et al. · NeurIPS 2026, AABA4ET Workshop · Project →
Generative Deepfake Videos in the Foundation-Model Era: A Timeline of Eroding Trust in Visual Evidence
Raza, S., Ho, J., Raza, M., Emmanouilidis, C. · ACM MAD '26 Workshop (2026) · DOI →
Just as Humans Need Vaccines, So Do Models: Model Immunization to Combat Falsehoods
Raza, S. et al. · IEEE WCCI / IJCNN 2026 · arXiv →
A Unified Evaluation Framework for Trustworthy LLMs, Agentic AI, and Multimodal Systems
Raza, S. et al. · 2026 · arXiv →
Harness-Aware Evaluation of LLM Agents
Raza, S. et al. · 2026 · Preprint →
Transparency in Agentic AI: A Survey of Interpretability, Explainability, and Governance
Raza, S. et al. · 2026 · Project →
FairLens: Fairness in Vision-Language Models
Raza, S. et al. · 2026 · arXiv →
UnBias-Plus: Detect, Explain, and Rewrite Bias
Raza, S. et al. · 2026 · arXiv →
Deepfakes in the Foundation-Model Era: A Survey of Forensics, Generation and Distribution Across the Social Media Lifecycle
Raza, S., Ho, J., Radwan, A. Y., Hafez, M. · 2026 · Preprint →
The Deepfakes We Missed: We Built Detectors for a Threat That Didn't Arrive
Raza, S. · 2026 · arXiv →
VLDBench: Evaluating Multimodal Disinformation with Regulatory Alignment
Raza, S. et al. · Information Fusion (2026)IF 18.6 · Q1 · DOI →
From Features to Actions: Explainability in Traditional and Agentic AI Systems
Chaduvula, S. et al. (incl. Raza, S.) · 2026 · arXiv →
F-DPO: Reducing Hallucinations in LLMs via Factuality-Aware Preference Learning
Raza, S. et al. · ACL 2026 Findings · ACL Anthology →
TRiSM for Agentic AI: Trust, Risk and Security in LLM-based Agentic Multi-Agent Systems
Raza, S. et al. · AI Open (2026) · DOI →
Review-based Recommender Systems: A Survey of Approaches, Challenges and Future Perspectives
Hasan, E., Rahman, M., Ding, C., Huang, J., Raza, S. · ACM Computing Surveys (2026) · DOI →
A Comprehensive Review of Recommender Systems: Transitioning from Theory to Practice
Raza, S. et al. · Computer Science Review (2026) · DOI →
Sustainable Open-Source AI Requires Tracking the Cumulative Footprint of Derivatives
Raza, S. et al. · ICML 2026 (Spotlight Position Paper) · arXiv →
Developing Safe and Responsible Large Language Models
Raza, S. et al. · Machine Learning, Springer (2025)IF 7.5 · Q1 · Paper →
Beyond Content: How Grammatical Gender Shapes Visual Representation in Text-to-Image Models
Raza, S. et al. · EMNLP Findings (2025) · arXiv →
RESPECT: A Framework for Promoting Inclusive and Respectful Conversations in Online Communications
Raza, S. et al. · Natural Language Processing Journal (2025) · DOI →
MBIAS: Mitigating Bias in Large Language Models While Retaining Context
Raza, S., Raval, A., Chatrath, V. · ACL 2024, WASSA Workshop · arXiv →
NBIAS: A Natural Language Processing Framework for Bias Identification in Text
Raza, S. et al. · Expert Systems with Applications (2023)IF 8.5 · Q1
Fake News Detection Based on News Content and Social Contexts: A Transformer-Based Approach
Raza, S. & Ding, C. · Int. J. Data Science & Analytics (2022)IF 2.4 · 200+ citations
Progress in Context-Aware Recommender Systems — An Overview
Raza, S. & Ding, C. · Computer Science Review (2019)IF 13.3 · Q1

Invited Talks & Keynotes

Accelerate Her Future Virtual Summit 2026
Panelist · Nov 20, 2026
Who benefits from AI? Who's left out? · Panel on building AI for good, centring Indigenous, Black, and racialized women
MLOps North Toronto Summit
TMLS · RBC Waterpark Place, Toronto · Nov 6, 2026
"Responsible AI Beyond Accuracy: Fairness, Safety, and Sustainability" · Evals & Benchmarks track Event →
MCP Dev Summit Toronto
Speaker · Oct 2026
Agentic AI evaluation, harness engineering, and responsible AI · building agents that are reliable, secure, and accountable
Future Technologies Conference 2026
SAI Conferences · Oct 2026
"From Features to Actions: Explainability in Traditional and Agentic AI Systems" · presented by Aravind Narayanan
HAICON 2026
Helmholtz Munich · Jun 2026
UNLOCK Initiative — "Current Status of the Benchmarking Field: Lessons Learned" · Vector–Helmholtz Munich partnership
ICML 2026
GREEN-AI Workshop · Helmholtz AI · Spotlight
"Sustainable Open-Source AI Requires Tracking the Cumulative Footprint of Derivatives" · Data & Impact Accounting (DIA) framework · Spotlight Position Paper
AAAI 2026
AI Governance Workshop · Singapore · Jan 2026
Panelist — AI Governance, Safety and Responsible AI · alongside IBM Research, Google DeepMind, Microsoft Research
IEEE SMC 2025
Keynote · Oct 2025
Responsible AI and Multimodal Fairness
NeurIPS 2025
AI Governance Workshop · Co-organizer
Co-organizer & Presenter — Policy, Risk, and Audit for Trustworthy AI
ResearchTrend.AI
Jun 2025
"HumaniBench: A Human-Centric Framework for Large Multimodal Models Evaluation"
Cohere, USA
Jul 2024
"Mitigating Biases in Text & LLMs — Challenges and Strategies"
Conestoga College
AI Symposium · Nov 2024
"AI and Data Privacy and Ethics"
ECIR 2024
Scotland · Mar 2024
"Dissecting News Narratives and Misinformation"
York University
Lassonde School · Mar 2024
"Enhancing Fairness in Large Language Models"
Rogers Catalyst
TMU · Jan 2024
"Cybersecurity and AI — Who Loses? The Case for More Inclusive Engagement"
AAAI 2023
Human-Medical AI Symposium · Aug 2023
"Fairness in Machine Learning Meets with Equity in Healthcare"
ICA 2023
Keynote · Jun 2023
"Accuracy Meets Diversity in a News Recommender System" · Keynote Address
Rice University
BRC Dept · Apr 2023
"Connecting Fairness in Machine Learning with Public Health Equity"

Professional Service

Program Committees

  • NeurIPS — Main Track + RegML, LAW, SoLaR, GenProCC, Position Paper Workshops
  • AAAI · ACL · ICLR · ICML · KDD
  • IJCAI · CIKM · ECCV · IEEE Big Data
  • AI Governance Workshop @ NeurIPS 2025 — Co-organizer
  • AAAI Workshop on Responsible LMs (ReLM 2024) — Co-lead

Editorial Roles

  • Springer Discover Computing — Editorial Member (2025–present)
  • Elsevier CS Portfolio — Transfer Manager & Scientific Editor (2021–present)
  • IEEE Transactions on AI — Review Editor (2023–present)
  • Frontiers in Big Data — Review Editor (2022–present)

Journal Reviewing

  • Nature · Nature Machine Intelligence
  • Lancet Digital Health · Science Robotics
  • JAMA Network Open · JAMIA
  • BMC Public Health · PLOS ONE
  • IEEE Trans. Knowledge & Data Engineering
  • Artificial Intelligence in Medicine

Grant Review Committees

  • CIHR Doctoral Research Awards — Peer Review Committee Member
  • National Killam Program — Peer Reviewer

Awards & Recognition

🏆
Responsible AI Leader of the YearWomen in AI Summit 2025 (North America)
⭐
The Peak Emerging Leaders 2026Artificial Intelligence · Canada
🔬
Top 2% Scientists WorldwideStanford / Elsevier 2025 & 2026 Updates
🎖️
ICML 2026 Spotlight Position PaperGREEN-AI Workshop, Helmholtz AI
🎓
Governor General Gold Medal NomineeToronto Metropolitan University, 2021

Grants & Fellowships

EU Horizon AIXPERT€515K
Co-Applicant · Grant No. 101214389 · 2025–present
CIHR Health System Impact Fellowship$175K
2021–2023
CIHR-IPPH Postdoctoral Research Fellowship$85K/yr
2021–2023
CIHR Post-COVID Disparity Analysis$100K
Co-Applicant · 2023–2024
AI4PH Postdoctoral Award$20K/yr
CIHR IRSC · 2022–2023
Ontario Graduate Scholarship$15K/yr
Twice awarded · 2019–2021
NSERC Research Stipend$11K/yr
4 years · 2016–2020

Technical Expertise

LLMs & Vision-Language Models
Responsible AI & Fairness
Green AI & Sustainability
AI Governance & Policy
Multimodal AI Benchmarking
NLP & Text Bias Detection
Health Informatics & Equity
PyTorch · HuggingFace · AWS · MLflow

Media Coverage

→ Metroland / InsideHalton: Deepfake election videos, a Toronto expert explains the red flags (Oct 2026) → SmarterArticles: Your Face Is Not Yours, the legal hole under deepfake law (Sep 2026) → Nosian Magazine: AI's Unasked Question, What Did That Cost? (Aug 2026) → BetaKit: Vector Institute launches open-source AI bias detector (Jul 2026) → SecurityBrief Canada: Vector launches open-source writing bias detector (Jul 2026) → Tech Talent Canada: Free AI tool to help HR and AI teams detect bias (Jul 2026) → Canadian Healthcare Technology: UnBias-Plus detects and rewrites bias in text (Jul 2026) → BornCity (DE): UnBias-Plus, Vector Institute startet kostenloses Bias-Erkennungs-Tool (Jul 2026) → AI Loop: Vector Institute unveils UnBias-Plus (Jul 2026) → GlobeNewswire: Vector Institute releases UnBias-Plus (Jun 2026) → Helmholtz Munich: Strengthening international AI research collaboration with Vector (Jun 2026) → AIwire: Vector Institute & Helmholtz Munich sign MOU (Jun 2026) → AIXPERT: Can AI see the world without stereotypes? (Jul 2026) → AIXPERT: Advancing Trustworthy, Explainable, and Responsible AI at NeurIPS 2025 (Jan 2026) → ISED Canada: Canadian Horizon Europe Secretariat newsletter feature (Mar 2026) → Vector Institute blog: When AI watches and listens, introducing SONIC-O1 (Jun 2026) → Vector Institute blog: Mixture of Experts, from sparse routing to multimodal deployment (2026) → Vector Institute 2025–26 Annual Report: Top 2% Scientists recognition → Vector Institute: UnBIAS Framework Revolutionizes Ethical Text Analysis