Muhammad Anser Sohaib

Hi, I'm Anser! I'm a Software Engineer at Hazen.ai developing production ML models deployed across highway and community surveillance systems internationally. My research focuses on efficient AI systems, particularly quantization techniques for edge deployment and intelligent resource optimization.

I investigate how to make advanced AI models truly deployable on edge devices through novel compression and security-aware optimization techniques. The challenge lies in universal accessibility and practical deployment across diverse hardware constraints.

I find inspiration in nature’s resilience, often seeing reflections of it in the AI systems I aim to design.

Email  /  LinkedIn  /  CV  /  Github

profile photo
★

Latest News

June 2025
Started as Associate Software Engineer at Hazen.ai, working on production ML models for surveillance systems.
July 2025
Three research papers submitted to top-tier conferences including NeurIPS 2025 and IEEE journals.
May 2025
Graduated with BSc Electrical Engineering from NUST with focus on AI and Machine Learning.
April 2025
Presented a Poster for my FYDP at Open House 2025
January 2022
Got tuition fee scholarship from NUST.
November 2021
Got admisison offer from NUST.
July 2021
Awarded a scholarship from ECAT.

Experience

Associate Software Engineer, Hazen.ai — Jun 2025 – Present

  • VAE-based latent extraction and reconstruction verification for license plate OCR.
  • OCR models for license number, country, and state; 32M+ synthetic images for robust training.
  • Optimized pipelines for real-time inference on Jetson.

ML Intern, SoC Lab, SINES, NUST — Aug 2024 – Mar 2025

  • GANs for high-fidelity face generation; improved robustness under masks and distortions.
  • Worked on IceWatch using transformers, CNNs, and LSTMs.

AI Intern, Global Resources — Jun 2024 – Aug 2024

  • Integrated LLMs into robotic automation; improved interfaces and logistics.

Publications/Preprints

A summary image for the publication titled 'TITAN: Transformer-Based Intelligent Task Allocation Network for Dual-Phase Learning and Optimal Scheduling'
TITAN: Transformer-Based Intelligent Task Allocation Network for Dual-Phase Learning and Optimal Scheduling
M. Anser, M. Talha, Z. Fatima
NeurIPS 2025 – Under Review
Abstract Paper Code
Study area for this paper
A Multimodal Framework for Glacial Environmental Analysis: Prospects for GLOF Risk Modeling
M. Anser, M. Talha, A. Kanwal, Z. Fatima, S. Sultana, N. Perwaiz
IEEE HONET-ICT 2025 – Submitted
Abstract Paper Dataset
A summary image for the publication titled 'IceWatch'
IceWatch: Forecasting Glacial Floods Using AI
M. Anser, Z. Fatima, M. Talha, A. Kanwal, N. Perwaiz, S. Sultana
IEEE JSTARS – Under Review
Abstract Paper Demo