Muhammad Ashar Javid

AI Researcher & Engineer specializing in 6G Wireless Networks and Multi-Agent Systems

Based in: Winnipeg, CanadaCurrently: Multimodal Prediction Researcher, Wireless Research Group, SEECS, NUSTExperience: 3 years

Hire me

Muhammad Ashar Javid is a Master's student at the University of Manitoba with a strong research background in RIS optimization, graph neural networks, and LLM-orchestrated resource allocation. He has authored multiple IEEE publications and held leadership roles, including CTO at E-Deviser and AI Engineer at Korea EHT.

Experience

Multimodal Prediction Researcher

Wireless Research Group, SEECS, NUSTIslamabad, PakistanworkNov 2025 – Present

Summary

Advisor: Dr. Neelma Naz

What I did

  • Curated and annotated DeepSense 6G scenarios for blockage prediction and adapted the dataset to support RIS-aided beamforming simulations.
  • Developed proactive blockage prediction and RIS reconfiguration methods for mobile vehicle-to-infrastructure links; modeled channel behavior using Rician fading.
DeepSense 6GRIS-aided beamformingRician fadingChannel modeling

AI for Wireless Networks Researcher

Information Processing and Transmission Lab, NUSTIslamabad, PakistanworkApr 2025 – Present

Summary

AI for Wireless Networks; Advisor: Dr. Syed Ali Hassan

What I did

  • Developed MAFRO, a multi-agent RIS optimization framework with a retrieval-augmented coordinator-evaluator, targeting sum rate, fairness and energy efficiency in dynamic wireless networks.
  • Generated optimal discrete phase-shift datasets through exhaustive search for language-model fine-tuning; benchmarked the framework against gradient descent, alternating optimization and manifold optimization.
  • Designed ACP-N for distributed agent coordination and RIS/NOMA resource allocation. These projects led to first-author publications in IEEE WCNC 2026 and IEEE Network.

Results

  • Achieved around 57% enhancement in a composite score of energy efficiency, SINR, and fairness compared to previous techniques.
PythonLLM fine-tuningRAGMulti-agent systemsRIS optimizationNOMAGradient descentManifold optimizationns-3

AI Engineer

Korea EHTcontractJun 2025 – Apr 2026

Summary

Part-time Contract

What I did

  • Built Smart Trace, an electric heat tracing automation platform using multimodal LLMs, reducing engineering design-cycle time by over 90%; the project was showcased at ADIPEC 2025.
  • Fine-tuned models for AutoCAD-based drawings and integrated retrieval-augmented memory for rule-consistent outputs and edge-case handling.

Results

  • Reduced engineering design-cycle time by over 90%.
  • Project showcased at ADIPEC 2025.
Multimodal LLMsAutoCADRAGModel fine-tuning

Skills

Technical

PythonPython
RIS optimization
LATEX
GNNs
MATLAB
PyTorchPyTorch
Transformers
LLM fine-tuning
RAG
Multi-agent systems
Resource allocation
Channel modeling
SciPy
LSTMs
Image classification
C/C++C/C++
GitGit
TensorFlowTensorFlow
scikit-learn
pandas
OpenCVOpenCV
BashBash
ns-3
SQLSQL
CVX
C++C++
ReduxRedux
TypeScriptTypeScript
ReactReact

Projects

Smart Trace

AI Engineerprofessional
An electric heat tracing automation platform using multimodal LLMs, reducing engineering design-cycle time by over 90%. Fine-tuned models for AutoCAD-based drawings and integrated retrieval-augmented memory.
Multimodal LLMsAutoCADRAGPython

Spherical Graph Neural Network for Seismic Propagation

Research Internacademic
Developed a spherical graph neural network combining Chebyshev spectral convolutions and bidirectional LSTMs to model seismic propagation for early earthquake detection. Evaluated on USGS and IRIS data.
GNNChebyshev spectral convolutionsBidirectional LSTMsPython

MAFRO

Lead Researcheracademic
A multi-agent RIS optimization framework with a retrieval-augmented coordinator-evaluator, targeting sum rate, fairness and energy efficiency in dynamic wireless networks.
Multi-agent systemsRAGPythonLLM fine-tuningns-3

Slingshot

personal
A goalkeeping-based game.
C++

Xplore Bot

personal
An AI-based travel booking assistant that automated the process of finding and booking holiday trips.
ReactRedux

Personal Portfolio

personal
TypeScript

Skin cancer classifier

personal
Built a skin cancer classifier during a research internship using CNNs and logistic regression with robust feature extraction preprocessing.
CNNsLogistic regression

Education

Master of Science

University of ManitobaElectrical Engineering (Thesis-based)2026

Thesis. decision-aware world models for decentralized mobility and resource control in multi-UAV networks

Bachelor of Engineering

National University of Sciences and Technology (NUST)Electrical EngineeringGPA 3.642022 – 2026

Thesis. Real-time RIS Optimization for 6G Networks Using a Multi-Agent Framework

Recognition

Awards

Mitacs Globalink Research Internship

University of Alberta, Canada2025

IPT Lab: Excellence in Innovative Research

Information Processing and Transmission Lab, NUST

NUST High Achiever's Gold Medal

National University of Sciences and Technology

NUST Merit Scholarship

National University of Sciences and Technology

Academic achievement

Patents & publications

Intelligent Multi-Agent Framework for RIS Optimization in 6G: A RAG-based Approach

publication2026

IEEE Wireless Communications and Networking Conference (WCNC)

With Muhammad Sameer Amjad, Muhammad Jamshaid Ghaffar, Haejoon Jung, Aamir Mahmood, Mikael Gidlund, Syed Ali Hassan

ACP-N: An Agentic Communication Protocol for Dynamic Resource Management in Next Generation Wireless Networks

publication2026

IEEE Network

With Muhammad Sameer Amjad, Muhammad Jamshaid Ghaffar, Haejoon Jung, Aamir Mahmood, Mikael Gidlund, Syed Ali Hassan

Semantics-to-Spectrum: LLM-Orchestrated Two-Timescale Resource Allocation for Cell-Free Massive MIMO in Industrial IoT Networks

research paper

Computer Communications; under review

With Abeha Hussain, Amna Akhtar Nawabi, Sohaib Amir Bukhari, Kapal Dev, Syed Ali Hassan

CA-GNN: A Constraint-Aware Graph Neural Framework for Max-Min Fair AP-UE Association in Cell-Free Massive MIMO

research paper

IEEE INFOCOM; under review

With Syed Muhammad Abdullah Rashid, Syed Ali Hassan

Leased Authority: Network-Issued Bounds on Agent Decision-Making for Heterogeneous Unmanned Systems in Low-Altitude Airspace

research paper

IEEE Wireless Communications; under review

With Amna Akhtar Nawabi, Sohaib Amir Bukhari, Abeha Hussain, Kapal Dev, Haejoon Jung, Alexandros-Apostolos A. Boulogeorgos, Syed Ali Hassan

Fine-Tuned Small Language Model for Adaptive Resource Optimization in RIS-Assisted ISAC Systems

research paper

IEEE GLOBECOM Workshops; submitted

With Abeha Hussain, Sohaib Amir Bukhari, Amna Akhtar Nawabi, Syed Ali Hassan

Using Quantized Small Language Models for RIS Optimization at the Edge in 6G Networks

research paper

IEEE International Conference on Communications (ICC); submitted

With Syed Mohammad Abdullah Rashid, Haejoon Jung, Syed Ali Hassan

When to Reconfigure an Intelligent Reflecting Surface: A Mobility-Aware Age-of-CSI Framework

research paper

IEEE International Conference on Communications (ICC); submitted

With Rabee Umar, Syed Ali Hassan

Social impact & activities

Poster Presentation

University of Alberta Research Symposiumconference talk2025

Presented work on spherical graph neural networks for seismic propagation.

Project Showcase

ADIPEC 2025other2025

Showcased Smart Trace, an electric heat tracing automation platform.

Hire MuhammadReach out about a role, a contract or a conversation.For recruiters

Muhammad's twin is AI, it can make mistakes.

More people on Kanvis

Browse more peopleCreate your own Kanvis