Poster presentation at the IEEE International Conference on Image Processing (ICIP) 2026 for YawDD+: Frame-level Annotations for Accurate Yawn Prediction.
About
I am a PhD Student at Silicon Austria Labs, GmbH and the University of Innsbruck, Austria, where my doctoral research centres on full model-heterogeneous federated learning methods for edge platforms. My work sits at the intersection of efficient machine learning, embedded systems, and distributed intelligence — spanning neural network quantization, pruning, knowledge distillation, and federated learning tailored to resource-constrained hardware. Prior to my PhD, I worked as an AI Researcher at DeltaX Co., Ltd. in Seoul, collaborating with Hyundai-Mobis and KIA on end-to-end deep learning pipelines for automotive applications including Driver Monitoring Systems and In-Cabin Monitoring Systems. This industry experience gave me hands-on expertise in hardware-software co-design and the deployment of optimised models on embedded platforms such as NVIDIA Jetson, Texas Instruments TDA4X, and HAILO. Earlier, as a Graduate Research Assistant at Information Security and Machine Learning Lab, Gachon University, I conducted foundational research on TinyML, publishing peer-reviewed work on CNN quantization and efficient inference on ARM Cortex-M microcontrollers.
Latest Updates
Poster presentation at the 28th International Conference on Pattern Recognition (ICPR) 2026 for FedADAS: Communication-Efficient Federated Distillation for On-Device Driver Yawn Recognition in Vehicular Networks.
Paper accepted at ICIP 2026 conference, YawDD+: Frame-level Annotations for Accurate Yawn Prediction.
Paper accepted at ICPR 2026 conference, FedADAS: Communication-Efficient Federated Distillation for On-Device Driver Yawn Recognition in Vehicular Networks.
Presentating my paper "Federated Distillation on Edge Devices: Efficient Client-Side Filtering for Non-IID Data" at FLTA 2025, Dubrovnik, Croatia.
Presentation on the topic "Collaborative Learning on the Edge" at the annual SAL-DC Summit, 2025.
Paper accepted at FLTA conference, Federated Distillation on Edge Devices: Efficient Client-Side Filtering for Non-IID Data.
Joined Silicon Austria Labs, GmbH as a PhD Student, Graz, Austria.
Joined DeltaX Co., Ltd. as an AI researcher, Seoul, South Korea.
Graduated in M.S Computer Engineering from Gachon University, Global Campus, South Korea.
Successfully defended my Masters thesis on, TinyML: Efficient Quantization and Deployment of Deep Neural Networks for Embedded IoT Applications, at Gachon University, Global Campus, South Korea.
Paper accepted at NeurIPS workshop, MLOps: Open Challenges from Hardware and Software Perspective in TinyML Devices.
Paper accepted at IEEE Transaction on Circuit and Systems II - Express Briefs, Low Latency Implementations of CNN for Resource-Constrained IoT Devices.
Joined as a Graduate Research Assistant in Information Security and Machine Learning Lab, and started M.S in Computer Engineering, Gachon university, South Korea
Graduated in B.Sc Electrical Engineering from University of Engineering and Technology, Taxila, Pakistan.
Research Interests
EdgeAI
Deploying intelligent models directly on edge devices with minimal latency and power consumption.
TinyML
Running machine learning inference on microcontrollers and ultra-constrained embedded systems.
Distributed Learning on Edge Devices
Model-Heterogeneous Federated Learning, Knowledge Distillation, EdgeAI
Experience
Ph.D. Student
Silicon Austria Labs GmbH · Graz, Austria
- Presented EdgeFD framework - A lightweight federated distillation framework for edge devices that beats SOTA frameworks.
- Further conducting doctoral research on updated EdgeFD versions such as knowledge distillation without pretrained teacher model under severe non-IID scenarios.
- Investigating federated distillation in automotive applications (DMS) and reinforcement learning.
- Motivated to publish at high-ranked international conferences and journals.
AI Researcher
DeltaX Co., Ltd. · Seoul, South Korea
- Delivered end-to-end deep learning solutions for automotive AI in collaboration with Hyundai-Mobis and KIA, including Driver Monitoring Systems (DMS) and In-Cabin Monitoring Systems (ICMS).
- Designed and deployed hardware-software co-optimised models on NVIDIA Jetson, TI TDA4X ADAS chips, and HAILO accelerators.
- Built complete application deployment pipelines for computer vision solutions on embedded edge platforms.
- Applied model compression techniques (quantization, pruning) to meet strict latency and memory budgets, and developed efficient inference pipelines using Python multi-threading and multiprocessing for real-time performance.
Graduate Research Assistant
ISML Lab, Gachon University · South Korea
- Researched TinyML with a focus on quantization and pruning of CNNs for image classification and object detection.
- Deployed deep neural networks on ARM Cortex-M processors using CMSIS-NN and CMix-NN libraries.
- Published peer-reviewed journal and conference papers on efficient DNN deployment for resource-constrained IoT devices.
- Developed and benchmarked optimised quantization libraries targeting the STM32H7 microcontroller.
Research Intern
QONCEPT LABS · Islamabad, Pakistan
- Surveyed state-of-the-art research in Natural Language Processing and Large Language Models.
- Compiled and presented a comprehensive literature review on ongoing LLM research directions.
Education
Oct 2025 – Present
Ph.D. in Computer Science
University of Innsbruck · Austria
Thesis: "Model-Heterogeneous Distributed Learning Methods for Heterogeneous Edge Computing Environments."
Supervisor: Gleb Radchenko
Co-supervisor: Radu Prodan
Mar 2021 – Feb 2023
M.S. Computer Engineering
Gachon University · Republic of Korea
Thesis: "TinyML: Efficient Quantization and Deployment of Deep Neural Networks for Embedded IoT Applications"
CGPA: 4.19 / 4.5 (93.25%) · Graduated with Best Paper Award
Supervisor: Seong Oun Hwang
Co-supervisor: Wai Kong Lee
Sept 2016 – Aug 2020
B.S. Electrical Engineering
University of Engineering & Technology (UET) · Taxila, Pakistan
Thesis: "Swarm Communication via Mobile Ad-hoc Networks (MANETs)"
CGPA: 3.53 / 4.0 (88.25%) · Graduated with Distinction
Supervisor: M. Danish Nisar
Publications
Journal Articles
Mujtaba, A., Lee, W.K., Ko, B.C. et al. “AuGQ: Augmented Quantization Granularity to Overcome Accuracy Degradation for Sub-Byte Quantized Deep Neural Networks.” Applied Intelligence, 55, 589, 2025.
Conference Papers
Rozi Khan, Ahmed Mujtaba, Wai Kong Lee, and Seong Oun Hwang. “Evaluation of Additive Power of Two Quantization on Various Deep Learning Datasets.” 9th International Conference on Green and Human Information Technology (ICGHIT), pp. 72–74, 2021.
Workshops
A. Mujtaba, Majeed, A., and S. O. Hwang. “MLOps: Open Challenges from Hardware and Software Perspective in TinyML Devices.” Challenges in Deploying and Monitoring ML Systems Workshop, NeurIPS, 2022.
Patent
Computing Device Performing Quantization of a Neural Network and Operating Method of the Computing Device
DOI: 10.8080/1020230032595Research Projects
CRYSTALLINE
Oct 2024 – Oct 2028Funded from Horizon Europe’s Excellent Science programme under the Marie Skłodowska-Curie COFUND Action, grant agreement No: 101126571.
eleCtRonic-based sYSTems reseArch exceLLence IN Europe
- For a duration of 4 years, with the expectation of gaining a PhD at the end of this period.
- Successful admission to the PhD program of the partner university and to the SAL-DC required.
- Conducting research in edge-oriented model-agnostic distributed learning under severe non-IID conditions.
EDISON
Sept 2025 – Aug 2028Austrian Research Promotion Agency (FFG)
Edge AI-based Decentralized Storage Optimization for Smart Energy Networks
- Conducting research on mulit-site federated learning for smart energy management at virtual power plants.
- Ablation study on existing federated learning algorithms for home energy management systems.
- Exploring the applicability of federated distillation for home energy management systems.
FAST AI Convergence
Mar 2021 – Feb 2023BK21 FOUR — Ministry of Education, Republic of Korea
Fast, Advanced, Secure and Transaction Artificial Intelligence Convergence
- Surveyed edge-based lightweight deep learning model deployment strategies.
- Proposed and experimentally validated novel optimisation techniques to reduce DNN inference latency on edge hardware.
- Produced peer-reviewed publications in premium international journals.
EQDL
May 2021 – May 2022Korean IITP International Collaboration Funding · Collaborator: University of Birmingham, UK
Efficient Quantization Techniques for Deep Learning in Embedded Systems
- Identified and characterised suitable microcontroller platforms for on-device DNN inference.
- Implemented and benchmarked a baseline quantization framework optimised for the STM32H7 MCU.
- Extended CMSIS-NN and CMix-NN libraries to better exploit STM32H7 hardware resources for accelerated inference, and published a peer-reviewed article in IEEE TCAS-II journal.
Skills
Programming
C / C++
Python
JavaScript
MATLAB
Verilog
AI Frameworks
TensorFlow
PyTorch
Hardware Platforms
NVIDIA Jetson
TI TDA4X
HAILO-15 SBC
STM32 MCUs
Arduino
ARM Cortex-M
Embedded ML / DL
CMSIS-NN
CMix-NN
STM32Cube.AI
Deployment & DevOps
Docker
Git
GitHub
Linux
Windows
Collaboration Tools
Confluence
JIRA
Honours & Awards
CRYSTALLINE
SAL Doctoral College
Oct 2024 - Oct 2028
Fully-funded doctoral scholarship under MSCA-COFUND Action, grant agreement No: 101126571.
Best Paper Award
Gachon University
February 2023
Awarded for publication in IEEE Transactions on Circuits and Systems II: Express Briefs.
Global Scholarship
Gachon University
Mar 2021 – Feb 2023
Fully-funded scholarship awarded by Gachon University, South Korea, for Master's in Computer Engineering.
Bachelor's Degree with Distinction
UET Taxila, Pakistan
Sept 2016 – Aug 2020
Graduated with distinction in Electrical Engineering with a CGPA of 3.53/4.0 (88.25%).
Ignite Grant (NGIRI)
National Technology Fund, Pakistan
Sept 2019 – Aug 2020
Received NGIRI funding from Ignite – National Technology Fund for the final-year project on Swarm Communication via MANETs.
Leadership & Service
Silicon Austria Labs, Doctoral College
Acting as the communication link between the SAL-DC Core Team and the SAL-DC student body.
Organizing regular meetings (once per month) together with the student body and communicating any issues or topics raised to the Core Team, thereby representing the student body within the Core Team.
Organizing an “All Hands Students Meeting” twice per year.
IEEE Sensors Council Student Branch Chapter
Gachon University · Seoul Section, Region 10
Organised and led invited talks, technical gatherings, mentoring activities, and networking programmes for the student branch community.
IEEE Student Branch
Gachon University · Seoul Section, Region 10
IEEE CASE Student Branch
Islamabad Section, Region 10 · General Secretary, Computer Science Society