CPU Performance Engineering
A hands-on reading path through CPU architecture, profiling and optimisation, with primary sources and runnable benchmarks.
Usamah Zaheer
Machine Learning Software Engineer
Cambridge, UK
Machine Learning Engineer with 4+ years of experience in deep learning, computer vision, and edge computing. Strong background in end-to-end ML development, model optimisation, and MLOps practices.
Usamah Zaheer is a Machine Learning Software Engineer at Arm, where he designs and optimises ML infrastructure for Arm architectures. He is pursuing an MS in Artificial Intelligence at the University of Texas at Austin and holds an MS in Embedded Systems from the University of Leicester. His work spans deep learning, computer vision, edge computing, robotics, and MLOps across companies including Arm, Dyson, and the University of Leicester.
Selected Projects
How to Make Your Model Fast
A free book on making ML models run inside real latency, memory, power and cost budgets, from rooflines and Arm silicon to LLMs, robots and agents.
Arm AGI CPU
Arm’s data center CPU for agentic AI, built on the Neoverse platform.
Dyson 360 Vis Nav
Autonomous robot vacuum with 360° vision, SLAM navigation and 26 sensors.
Book
How to Make Your Model Fast
A Systems View of Efficient Machine Learning, from Silicon to Agents
A book in fourteen parts on making models run inside real budgets for latency, memory, power and cost. Rooflines and Arm vector units, kernels and compilers, quantisation and compression, then vision, on-device language models, robotics, profiling, serving and agents. Free to read at ai.usamah.me.
Experience
Machine Learning Engineer
Arm
- Designed and optimised ML infrastructure to analyse and enhance performance of models and systems on Arm architectures.
- Optimised ML compilers and libraries for Arm, improving inference performance and reducing latency.
- Conducted deep kernel-level analysis to identify and eliminate inference bottlenecks.
- Profiled and optimised runtime performance of ML models; developed scalable benchmarking solutions across cloud and edge environments.
- Built automated pipelines for data collection, preprocessing, and model evaluation, streamlining production workflows.
- Led cross-functional collaborations to align ML infrastructure with organisational goals, owning a new project from inception.
PyTorch · TensorFlow · JAX · FBGEMM · KleidiAI · ACL · ArmNN · OneDNN
Robotics Software Engineer
Dyson
- Developed CNN algorithms for segmentation, object detection, and classification, applying quantisation, pruning, and knowledge distillation for deployment on robot hardware.
- Architected evaluation tools and robotics algorithms for planning and navigation using C++. Conducted log analysis, debugging, and on-robot testing.
- Deployed models on diverse hardware and edge devices, optimising through profiling and bottleneck analysis using CUDA and cuDNN.
- Developed a VLM solution that saved over £100,000 and boosted productivity by 20x.
- Streamlined the ML lifecycle with model versioning, monitoring, and automated deployment using CI/CD pipelines.
- Presented complex ML projects to senior leaders and the CEO.
PyTorch · MXNet · ONNX · CUDA · cuDNN · C++
ML Research Assistant
University of Leicester
- Spearheaded development of an end-to-end automated ML pipeline for processing high-resolution data in real-time.
- Integrated cutting-edge CNNs in PyTorch and TensorFlow for high-resolution satellite imagery analysis.
- Led development of ML systems utilising Random Forest and SVM for predictive modelling.
- Deployed AI solutions in cloud environments with Docker and Kubernetes.
PyTorch · TensorFlow · Scikit-learn · Docker · Kubernetes · R
Projects
- Curated a reading path through x86 and Arm performance engineering, from individual instructions to production inference.
- Pairs primary sources with runnable C benchmarks and an MCP server for source-grounded performance research.
Arm Learning Paths
Arm
- Authored four hands-on Arm Learning Paths covering image super-resolution, pet classification, prompt segmentation and voice activity detection.
- Guides developers through model export, deployment and validation with ExecuTorch, Arm VGF and Ethos-U85.
ML Model Deployment App
- Developed an Android application for deploying ML models on edge devices for object detection using YOLO, Mask R-CNN, and SSD.
- Optimised using TensorFlow Lite with quantisation and pruning for low-latency, on-device inference.
360 Vision Navigation
Dyson
- Developed a robot utilising 26 sensors with SLAM technology and 360-degree vision for autonomous navigation.
- Advanced path planning algorithms and developed integration and unit tests in C++ and Python.
Air Purifiers Embedded Software
Dyson
- Contributed to embedded software for all Dyson air purifiers including Pure Cool using C and Python.
- Designed backbone logic behaviour system for hardware communication with FreeRTOS. Updated a fundamental library stack for 10+ projects.
AI4EO & CNN Research
University of Leicester
- Classified high-resolution satellite images for forest fire detection using CNNs, Random Forest, and SVM.
- Evaluated CNN architectures for autonomous vehicle applications, optimised with Transfer Learning and TensorRT.
AI Agent Systems
Stealth Startup
- Led design and implementation of AI agents for SDRs, integrating NLP and LLMs for accurate and scalable solutions.
- Managed cloud infrastructure on Vertex AI with Databricks and Snowflake, utilising RAG to enhance AI response precision.
- Orchestrated ML workflows and semantic search using LangChain and LlamaIndex.
- Oversaw containerised deployment with Docker and Kubernetes, implementing MLOps practices with MLflow.
Education
- MS in Artificial IntelligenceJan 2025 – Dec 2026
University of Texas at Austin, USA
- MS in Embedded Systems and Control EngineeringJan 2021 – Aug 2022
University of Leicester, England - Distinction
Thesis: Performance Evaluation of Deep Learning Techniques for Object Detection in Autonomous Vehicles
- BTech in Electronics and Communication EngineeringAug 2016 – Oct 2020
Jawaharlal Nehru Technological University, India - First Class
Skills
Languages: Python · C++ · C · Rust · SQL · MATLAB
ML & Deep Learning: PyTorch · TensorFlow · JAX · Scikit-learn · LLMs · VLMs · Multimodal Models · CNNs · Transformers
Computing & Inference: TensorRT · CUDA · cuDNN · ArmNN · LLVM · GGML · OpenMP · ONNX/Runtime · FBGEMM · KleidiAI
Profiling & Debugging: NVIDIA Nsight · PyTorch Profiler · TensorFlow Profiler · Valgrind · gprof · cProfile · Py-Spy
Cloud & Deployment: AWS · GCP · Vertex AI · Kubernetes · Docker · GitHub Actions · Jenkins · MLflow · KubeFlow · Databricks · Snowflake
Data & Visualisation: Pandas · Apache Spark · Matplotlib · Plotly · Streamlit · Grafana · Tableau · Gradio
Leadership
- Led an entire project from inception to completion independently at Arm.
- Presented projects to the Dyson CEO and senior leadership.
- Mentored undergraduates in robotics and machine learning.
- Participated in 10+ hackathons and workshops.
- Open-source contributor with a portfolio of projects on GitHub.
Contact
Latest Posts
Don't Be an Em Dash
Usamah Zaheer on getting replies from people he knows that were pasted out of ChatGPT and sent as their own, when a couple of words they typed themselves would have done.
We Can All Bake Bread
Usamah Zaheer on why AI letting anyone build their own software will not kill software companies. Everyone can bake bread and almost everyone still buys it, because nobody wants to be the baker.
AI Is Making Everyone Sound the Same
Usamah Zaheer argues that when everyone runs their thinking through the same models, outputs converge, and the scarce skill shifts from generating an answer to knowing whether it is any good.
Stop Latency Laundering
Usamah Zaheer names latency laundering, the habit of moving delay outside the measurement boundary until a slow system reports a fast number.
The Next Trillion-Dollar AI Opportunity Is Not Bigger Models
Usamah Zaheer argues the next trillion-dollar AI opportunity is not another model lab. It is the infrastructure that makes intelligence cheap enough to run on billions of devices.
Inside the Black Box: ML Compiler Optimisation from a Practitioner at Arm
Usamah Zaheer explains ML compiler optimisation from a practitioner perspective at Arm, covering graph optimisation, kernel selection, and the profiling feedback loop.
Why Edge ML Inference is the Next Frontier
Usamah Zaheer explains why on-device ML inference is the future, drawing on experience optimising models at Arm and deploying perception systems at Dyson.
I Gave a Robot Eyes and a Brain: VLMs in Real-World Robotics
The gap between a VLM demo on Twitter and a VLM running on a real robot is enormous. Usamah Zaheer shares what he learned closing that gap at Dyson.
Applying CNNs to Satellite Imagery: Lessons from Forest Fire Detection
Usamah Zaheer shares lessons from applying CNNs to high-resolution satellite imagery for forest fire detection at the University of Leicester.
What I Learned Building AI Agents at a Stealth Startup
Usamah Zaheer built AI agent systems from scratch at a stealth startup. Here is what the hype gets right, what it gets wrong, and why he chose to go back to school.
Life
An AI video I made with Higgsfield, with me as the main character.