Building AI systems that see, understand, and explain themselves.
I'm a researcher and engineer working at the frontier of Explainable AI (XAI) and Video Understanding. My mission is straightforward: AI systems that make consequential decisions should be able to tell us why.
I hold a PhD in Computer Engineering from Concordia University (Montreal), where I developed novel methods for interpreting video Transformer models, exposing their vulnerabilities, and deploying explainability at scale. I now work as a Core AI Engineer at Maket Technologies, leading explainable AI systems for generative AI applications.
My research spans three pillars:
Interpret — How do video Transformers actually make decisions? (STAA)
Stress-test — Can we break them to understand their limits? (Adversarial Attacks)
Operationalize — How do we ship explainability into production? (XAIport / XAIpipeline)
|
IEEE Access 2025 A single-forward-pass method that produces spatio-temporal explanations for video Transformers with <3% overhead. Outperforms Grad-CAM and Attention Rollout with 0.87 faithfulness on Kinetics-400.
|
ICSE 2024 A microservice framework that shifts explainability from post-hoc afterthought to an integral development practice. Works across Azure, GCP, and AWS out of the box.
|
|
ACM TOMM 2025 First joint spatio-temporal adversarial attack framework targeting video Transformer self-attention. Achieves state-of-the-art attack success rate on Kinetics-400, revealing systematic security vulnerabilities.
|
IEEE Trans. Cloud Computing 2024 An open API architecture for explaining proprietary cloud AI services (Azure, GCP, AWS) without accessing model internals. Full provenance tracking for reproducibility.
|
|
XAI-Service — Full-stack XAI platform combining XAIport + XAIpipeline. Cloud-agnostic explainability services with REST APIs, automated workflow orchestration, and CI/CD deployment. If you work with AI on the cloud, this is for you. |
12 peer-reviewed papers | 101+ citations | h-index: 5 | 3-year career span (2022-2025)
| Venue | Type | Rank | Papers |
|---|---|---|---|
| ICSE | Conference | CORE A*, h5: 74 | 1 |
| IEEE Trans. Cloud Computing | Journal | Q1, IF: 5.95 | 1 |
| ACM Trans. Multimedia | Journal | Q1, IF: 6.0 | 1 |
| IEEE Access | Journal | Q1, IF: 3.6 | 1 |
| IEEE SSE / COMPSAC / Big Data | Conference | IEEE flagship | 6 |
| IEEE Computer Magazine | Magazine | Flagship (co-author) | 1 |
| CSCE | Conference | Intl. | 1 |
- Peer Reviewer: 35+ manuscript reviews for IEEE Transactions on Services Computing, Applied Intelligence, AAAI, IJCNN, and more
- Recognized Reviewer: Springer Nature Official Certificate; AAAI Workshop personal acknowledgement
- Workshop Facilitator: CASCON 2024 — "Develop Explainable AI Services on Cloud Computing and Open Source Models"
- Teaching Assistant: Graduate courses in Software Engineering, Cloud Computing, and Distributed Systems at Concordia University
- Professional Member: IEEE Computer Society • ACM
Python PyTorch TensorFlow Transformers FastAPI Docker Kubernetes
Azure GCP AWS HuggingFace LangChain RAG MLflow
React Next.js Node.js PostgreSQL MongoDB Redis Git
| Degree | Institution | Focus |
|---|---|---|
| PhD, Computer Engineering | Concordia University, Canada | Explainable AI, Video Understanding, Transformer Models |
| MSc, Process System Engineering | TU Dortmund, Germany | Advanced Modelling, Distributed Systems |
| BSc, Process System Engineering | CUMT, China | Foundation Engineering |
Also: PMP Certified (PMI #2256006)
"The measure of intelligence is not whether a machine can think — but whether it can explain its thinking."
If you're working on interpretable AI, trustworthy ML, or video understanding — let's talk.
