CV
Summary
Autonomous-driving perception and end-to-end driving engineer working camera-based deep-learning perception since 2019, progressing from Perception Engineer to Senior Staff Engineer & Tech Lead. Hands-on across the full perception stack — 3D object detection, environment modeling, and online HD-map construction — with deep expertise in BEV architectures and Transformers, now leading end-to-end (vision-language-action) autonomous driving. PhD in Electrical Engineering (Best Dissertation Award).
Experience
Senior Staff Engineer & Tech Lead — Autonomous Driving
Progressed Senior Engineer → Staff Engineer → Senior Staff Engineer (Jan 2026).
- Developed camera-based 3D object detection across the full set of road agents — vehicles, vulnerable road users (pedestrians and two-wheelers), and traffic elements (signs and lights).
- Built static-world perception and environment modeling — lane and road detection, drivable-space estimation, road-boundary detection, and online HD-map construction — to enable driving without reliance on HD maps, generalizing across diverse ODDs from complex intersections to rural roads and multi-lane highways.
- Deep expertise across BEV (bird's-eye-view) perception paradigms — implicit, explicit, and geometric BEV — and Transformer-based detection, evolved across multiple model generations.
- Tech lead for end-to-end autonomous driving, directing vision-action (VA) and vision-language-action (VLA) models that map raw camera input to driving behavior.
- Core contributor to GeRo and MAPLE — latent world-model–augmented VLA models that rehearse future driving scenarios to improve robustness of end-to-end planning under rare and extreme cases.
Perception Engineer → Senior Perception Engineer
- Developed camera-based deep-learning object detection for autonomous-driving vehicles, meeting automotive robustness and real-time constraints.
- Delivered detection models that fed the perception stack later carried into Qualcomm's autonomous-driving program.
Postdoctoral Research Fellow
- Researched security and robustness of deep learning with Prof. Erik G. Larsson; authored the pioneering work on adversarial attacks against DL-based signal classification.
Selected Publications
- M. Sadeghi et al., “MAPLE: Latent Multi-Agent Play for End-to-End Autonomous Driving,” arXiv preprint, 2026. [paper]
- M. Sadeghi et al., “Generative Scenario Rollouts for End-to-End Autonomous Driving (GeRo),” arXiv preprint, 2026. [paper]
- M. Sadeghi and E. G. Larsson, “Adversarial Attacks on Deep-Learning Based Radio Signal Classification,” IEEE Wireless Communications Letters, 2018. [paper]
Awards & Honors
- Qualcomm IP Achievement Award, 2025 — awarded to ~50 of ~50,000 employees (top ~0.1%) for exceptional quality and impact of ideas and IP.
- Best PhD Dissertation Award, Singapore University of Technology and Design (SUTD), 2019.
- FIRST Industry Workshop Outstanding Graduate Research Award, 2017; MediaTek Graduate Research Competition Award, 2017.
- Merlion PhD Award, 2014; SUTD President's Graduate Fellowship, 2013; Singapore Graduate Fellowship, 2013.
Technical Skills
- Languages: Python, C++, MATLAB; HTML, CSS, JavaScript.
- Deep learning: PyTorch, TensorFlow/Keras; Transformers, BEV perception, world models, VLA / multimodal LLMs, generative models.
- Autonomous driving: online mapping, lane-graph construction, road & lane detection, object detection, environment modeling, end-to-end driving, data-driven development.
- Methods & tools: computer vision, optimization, adversarial robustness, Git, LaTeX.
Education
PhD, Electrical Engineering
Best PhD Dissertation Award. Advisor: Assoc. Prof. Chau Yuen.
MSc, Electrical Engineering
BSc, Electrical Engineering
Professional Service
- Reviewer: IEEE Transactions on Wireless Communications, Signal Processing, Communications, and Vehicular Technology; IEEE Signal Processing / Communications / Wireless Communications Letters.
- Technical Program Committee: IEEE ICCS 2018; IEEE VTS APWCS 2017–2018. Former officer, IEEE Emerging Technology Initiative on Machine Learning for Communications.