Omer Tariq
Ph.D. Machine Learning, KAIST
omertariq2000@gmail.com
Daejeon, Republic of Korea
KAIST, South Korea
Monitra, United Kingdom
Research Interests
Foundation Models
Parameter-Efficient Fine-Tuning, Retrieval-Augmented Generation, In-Context Learning, Chain-of-Thought Reasoning, Vision-Language Models (CLIP), Diffusion Models, Transformers
Navigation & Localization
Neural Inertial Navigation, Visual-Inertial Odometry, End-to-End SLAM, Multi-Modal Sensor Fusion, Secure Localization, Deep Reinforcement Learning, Self-Supervised Learning
Hardware-Aware AI
Neural Architecture Search, Quantization, Knowledge Distillation, AutoML, Hardware-Aware Optimization, Adversarial Robustness, Differential Privacy, Federated & Distributed Learning, FPGA/SoC Acceleration
Embodied AI
Vision-Language-Action Models (PaLM-E), Multimodal Perception and Control, Sim-to-Real Transfer
Zero-Shot and Few-Shot Learning
Meta-Learning, Prompt Engineering, Cross-Modal Transfer, Low-Resource Adaptation, Few-Shot Prompting Techniques
Education
Korea Advanced Institute of Science and Technology
MS & Ph.D., School of Computing
Thesis: Robust Domain-Invariant Inertial Localization in Real-Time Federated Edge Computing
CGPA: 3.73/4.3 (94%)
November 2021 - June 2025)
Advisor: Prof. Dongsoo Han
University of Engineering and Technology, Taxila
B.S., Electrical Engineering
Thesis: Realtime Object Detection and Autonomous Control using 3-DoF Robotic Arm
November 2010 - July 2014
Selected Publications
NanoMST: A Hardware-Aware Multiscale Transformer Network for TinyML-Based Real-Time Inertial Motion Tracking
2025Omer Tariq, Dongsoo Han
IEEE Internet of Things Journal
NanoMST is an efficient multi-scale transformer for inertial motion tracking that achieves high-precision trajectory estimation with minimal computational cost and real-time performance on edge devices, outperforming larger models across benchmarks while requiring only 298K parameters and supporting 8-bit quantization.
DeepILS: Towards Accurate Domain Invariant AIoT-enabled Inertial Localization System
2025Omer Tariq, B. Dastagir, M. Bilal, Dongsoo Han
IEEE Internet of Things Journal
This paper presents DeepILS, a domain-invariant AIoT-enabled inertial localization system that achieves high accuracy across diverse environments without requiring environment-specific retraining.
Meta-Swin: Lightweight Image Super-Resolution Swin Transformer for Metaverse Applications
2024Omer Tariq, Dastagir M.B.A., M. Bilal, Dongsoo Han
2nd International Conference on Intelligent Metaverse Technologies & Applications (iMETA)
This paper proposes Meta-Swin, a lightweight Swin Transformer architecture for image super-resolution in metaverse environments with constrained computational resources.
Recent Updates
Thrilled to announce my latest publication in the IEEE Internet of Things Journal (Impact Factor: 8.9)! 🚀 🔬 NanoMST: A Hardware-Aware Multiscale Transformer Network for TinyML-Based Real-Time Inertial Motion Tracking
July 04, 2025Started as a Machine Learning Engineer at Monitra, UK
March 01, 2025Started as a Teaching Assistant for Special Topics in Computer Science (CS492) at KAIST
March 01, 2025Our paper "DeepILS: Towards Accurate Domain Invariant AIoT-enabled Inertial Localization System" accepted to IEEE Internet of Things Journal Featured
February 15, 2025Our paper 'Meta-Swin: Lightweight Image Super-Resolution Swin Transformer for Metaverse Applications' accepted to iMETA 2024
November 10, 2024Current Positions
Machine Learning Engineer
Monitra, United Kingdom
March 2025 - Present
- Develop secure Python scripts to authenticate with Azure REST APIs and automate data ingestion from the Atlas platform
- Train and export machine learning models using TensorFlow for fault type classification in PRPD images
- Implement MLflow for experiment tracking and model versioning with deployment-ready architecture in CI/CD workflows
PhD Candidate
Intelligent Service Integration Lab, KAIST
October 2022 - Present
- Develop AIoT-enabled inertial localization systems for accurate, privacy-preserving navigation
- Design NanoMST: an ultra-lightweight multiscale transformer for TinyML-based motion tracking
- Implement real-time FPGA-based particle filter SLAM accelerator for mobile robot navigation
Honors & Awards
-
KAIST MS/Ph.D. Scholarship (20,000 USD/year)2021
-
Distinguished Service Award, SUPARCO2018
-
Distinguished Performance Award for PakTes-1A satellite launch2018
Volunteer Activities
Reviewer (International Journals)
- IEEE Internet of Things Journal
- IEEE Internet of Things Letter
- NUCLEAR SCIENCE AND TECHNIQUES (NST), Springer
- Engineering Applications of Artificial Intelligence, ScienceDirect