cv
AI/ML engineer — machine learning systems, reproducible research software, and research at QMUL.
Basics
| Name | Ammar Yasir Naich |
| Label | AI/ML Engineer |
| a.y.naich@qmul.ac.uk | |
| Phone | +44-7436792873 |
| Url | https://ammaryasirnaich.github.io |
| Summary | AI/ML engineer with a PhD in Artificial Intelligence and hands-on experience building efficient machine learning systems, reproducible research software, and scalable data pipelines for data-intensive and real-world applications. Comfortable across Python, distributed computing, GPU/CUDA optimisation, transformer-based models, quantization, and edge-oriented AI, with a track record of rigorous experimentation and delivery. Background spans industry engineering, academic research, teaching, and postgraduate supervision. |
Work
-
2025.09 - 2025.10 London, UK
Machine Learning Engineer
Digital Reality Corp (DRC)
- Developed and trained models converting LiDAR point clouds into 2D/3D digital assets.
- Optimised models and automated data pipelines with CI/CD and AWS services for scalable, reliable delivery.
-
2021.10 - Present London, UK
Research Fellow
Queen Mary University of London
- Nominated by QMUL to lecture at QUPT (Hainan, China) on the module Introduction to Data Science.
- Leading development of a scalable AI platform for ML model hosting and evaluation on OpenStack, with reproducible workflows and distributed training/deployment across GPUs; built internal tooling and CLI-based interfaces.
- Developed coursework and laboratory exercises for Big Data Processing (Apache Spark Streaming, Apache GraphFrames), managed lab delivery, and assessed student work.
- Supervised 23 MSc students in deep learning and computer vision projects.
- Designed and delivered advanced coursework and labs in Principles of Machine Learning, integrating crowdsourced datasets and improving student competency in ML techniques by 75%.
-
2020.01 - 2021.01 London, UK
Embedded Software Engineer
NodeNS
- Built a sensor integration unit for plug-and-play connections between mmWave radar sensors and edge devices.
- Designed and implemented a security protocol for secure communication across the networked system.
- Defined the development stack and built a GUI tool for sensor data transfer and configuration.
-
2019.06 - 2024.07 London, UK
PhD Research
Queen Mary University of London
- Developed and evaluated deep learning models for real-time 3D object detection using KITTI, nuScenes, and Waymo; built reproducible training and evaluation pipelines.
- Implemented custom CUDA kernels and leveraged model, data, and pipeline parallelism for efficiency across local and cluster GPUs.
- Explored quantization strategies for efficient training and fine-tuning of LLMs under edge and low-memory GPU conditions.
-
2011.01 - 2018.09 Pakistan
Technical Manager / Software Architecture
Stingray Technologies (Pvt) Ltd (formerly Maritime System Limited)
- Led software architecture and cross-functional teams delivering scalable, low-latency systems in C/C++/Qt.
- Developed software for real-time acquisition and processing from digital I/O and serial sensors; improved processing speed by 40%.
- Architected emulators and simulators in C/C++ for hardware testing, improving reliability and reducing project cost by 35%.
- Contributed to Critical Event Management Services across 250 possible system events, reducing downtime by 70%.
Education
-
2019.06 - 2024.07 London, UK
PhD
Queen Mary University of London
Computer Science (PhD thesis: LiDAR/transformer architectures and multi-GPU optimisation for 3D object detection)
-
Jamshoro, Pakistan
-
Jamshoro, Pakistan
Publications
-
2024 LiDAR-based intensity-aware outdoor 3D object detection
Sensors 24.9 (2024): 2942, MDPI
Awards
- 2025.01.01
- 2019.06.01
Volunteer
-
2022.01 - Present -
2022.01 - Present
Certificates
| Fundamentals of Accelerated Computing with CUDA Python | ||
| NVIDIA DLI | 2022-01-24 |
| Visual Perception for Self-Driving Cars | ||
| University of Toronto (Coursera) | 2021-09-01 |
| Big Data Specialization (Version 1) | ||
| University of California San Diego (Coursera) | 2016-11-17 |
Skills
| Machine learning research | |
| Transformers | |
| Deep learning | |
| LLM/SLM adaptation | |
| Quantization-aware training | |
| Post-training quantization | |
| Computer vision | |
| Multimodal learning |
| AI-assisted development | |
| Cursor | |
| OpenClaw | |
| LLM-based code generation and debugging | |
| Rapid prototyping | |
| Workflow automation |
| Programming and research software | |
| Python | |
| C++ | |
| SQL | |
| CUDA | |
| Reproducible experimentation | |
| Evaluation workflows | |
| Data processing |
| Frameworks and tooling | |
| PyTorch | |
| Hugging Face | |
| OpenMMLab | |
| TensorBoard | |
| LangChain |
| Large-scale computing | |
| Multi-GPU training | |
| Distributed training | |
| Apache Spark | |
| Spark Streaming | |
| Kafka | |
| GraphFrames |
| Infrastructure and systems | |
| OpenStack | |
| AWS | |
| Kubernetes | |
| Kubeflow | |
| CI/CD | |
| Git |
| Optimisation and efficient AI | |
| GPU/CUDA optimisation | |
| Low-latency inference | |
| Model compression | |
| Edge-oriented AI | |
| Efficient execution on constrained systems |
Languages
| Urdu | |
| Native speaker |
| Sindhi | |
| Native speaker |
| English | |
| Fluent |
Interests
| Computer vision | |
| 3D object detection | |
| LiDAR | |
| Vision and convolution transformers |
| Efficient and edge AI | |
| Quantization | |
| LLM/SLM on constrained hardware | |
| CUDA |
| Distributed ML | |
| Multi-GPU | |
| Spark | |
| Reproducible pipelines |
References
| Senior Lecturer Dr Jesús Requena Carrión | |
| j.requena@qmul.ac.uk |
| Senior Lecturer Dr Mona Jaber | |
| m.jaber@qmul.ac.uk |