Postgraduate Diploma in Accelerator-Aware Architectures and Performance Measurement
Online Postgraduate Diploma of 800 h in Accelerator-Aware Architectures and Performance Measurement. 8 modules with ISO 26262:2018, ISO/PAS 8800:2024 and NVIDIA Nsight Compute. Assigned tutor and open enrolment.
Postgraduate Diploma in Accelerator-Aware Architectures and Performance Measurement — studying from Australia
Australia has active research in autonomous systems, particularly in mining, agriculture and freight, where autonomous haulage and automated equipment are well established. Universities and research bodies collaborate closely with industry on perception, sensor fusion and vehicle software. Advanced driver assistance systems are increasingly standard in imported vehicles, creating demand for engineers who understand safety validation and software integration. Employers range from mining technology firms and research institutions to global automotive software suppliers operating local engineering teams.
Fees for Australia: A$2,380 (AUD), with instalments and employer sponsorship available.
Relevant bodies in Australia
Typical roles
Questions from Australia
Is this qualification recognised by Australian engineering bodies?
SEIUM's qualifications are private and not official state degrees, and they are not accredited by Engineers Australia or any other national body. Professional recognition for engineers in Australia usually runs through accredited university degrees and Engineers Australia membership. Whether a private qualification is valued is decided by employers or the relevant body.
Which Australian employers work in autonomous and ADAS software?
Australia has automotive suppliers, mining and logistics companies using autonomous systems, and technology firms working on driver assistance and vehicle software. Employers usually assess your coding ability, project portfolio and experience more than the name of a private course. Ask prospective employers how they regard the qualification.
How demanding is the course and is it workable with a job in Australia?
The programmes are online and taught in English, so you can study from home without needing a student visa. Specialist courses are 300 hours, postgraduate diplomas 800 hours and master's 1,900 hours, so consider your available time carefully. An assigned technical tutor supports you, fees are payable in instalments, and employer sponsorship may be possible.
Take the controls.
Run a real engineering trade-off from this field. Move the sliders, watch the model react in real time and beat the challenge.
What you will be able to do
Compare accelerator architectures for embedded AI.
Design neural networks adapted to hardware.
Use inference compilers and runtimes.
Apply a rigorous performance measurement methodology.
Profile and eliminate bottlenecks.
Carry out fair comparisons between platforms.
Integrate accelerators into system architecture with safety criteria.
Who it's for
- AI performance engineer who wants to specialise in Accelerator-Aware Architectures and Performance Measurement.
- AI hardware architect who wants to specialise in Accelerator-Aware Architectures and Performance Measurement.
- Compiler engineer who wants to specialise in Accelerator-Aware Architectures and Performance Measurement.
- Embedded machine learning engineer who wants to specialise in Accelerator-Aware Architectures and Performance Measurement.
- Benchmarking engineer who wants to specialise in Accelerator-Aware Architectures and Performance Measurement.
- ADAS systems engineer who wants to specialise in Accelerator-Aware Architectures and Performance Measurement.
- Graduates in engineering or sciences seeking to enter the automotive sector.
What you'll work with
- ISO 26262:2018
- ISO/PAS 8800:2024
- AEC-Q100
- JEDEC JESD89A
- ISO 21448:2022 (SOTIF)
- MLPerf Automotive
- IEC 61508:2010
8 modules · 800 h
1.1GPU, NPU, FPGA and ASIC1.2Systolic arrays1.3Tensor units and reduced precision1.4On-chip memory and data reuse1.5Roofline model1.6TOPS, TOPS/W and real utilisation1.7Critical reading of datasheets1.8Vendor roadmaps1.9Weight-stationary and output-stationary dataflows1.10Case study: comparison of commercial accelerators
2.1Operators supported by the hardware2.2Constrained architecture search2.3Efficient network families2.4BEV backbones2.5Efficient attention in transformers2.6Quantisation during training2.7Structured sparsity2.8Layer compatibility2.9Per-layer latency predictors2.10Laboratory: architecture search with a latency limit
3.1Intermediate representations and ONNX3.2TensorRT, TVM and XLA3.3Operator fusion3.4Memory planning3.5INT8 calibration3.6Vendor runtimes3.7Numerical discrepancies3.8Portability across platforms3.9Kernel autotuning3.10Practice: compiling a model with Apache TVM
4.1Latency, throughput and jitter4.2Cold and warm measurement4.3Percentiles and distribution tails4.4End-to-end versus per-layer4.5Reproducible conditions4.6Statistics of measurements4.7Hardware counters4.8Common benchmarking errors4.9Frequency and temperature control during measurement4.10Laboratory: reproducible measurement bench
5.1GPU profilers5.2Occupancy and memory5.3CPU-accelerator transfers5.4Pre-processing as a bottleneck5.5Applied roofline5.6Complete system traces5.7Improvement opportunities5.8Profiling reports5.9Shared DRAM bandwidth5.10Practice: roofline analysis of a detector
6.1MLPerf Inference and MLPerf Automotive6.2Perception benchmarks6.3Internal benchmarks6.4Fair comparison across platforms6.5Accuracy versus performance6.6Measured power consumption6.7Presentation of results6.8Continuous performance regression6.9Total cost per inference6.10Case study: comparative report on two platforms
7.1Partitioning the stack across accelerators7.2Redundancy in AI computing7.3Functional safety of accelerators7.4Transient errors7.5Soft errors and ECC7.6AEC-Q100 qualification7.7Performance evidence7.8Hardware lifecycle7.9Dual execution with comparison7.10Workshop: computing architecture for an L3 ADAS
8.1Application requirements8.2Chosen accelerator8.3Network adaptation8.4Compilation and deployment8.5Measurement plan8.6Profiling and optimisation8.7Comparison of results8.8Architecture recommendations8.9Report and presentation8.10Final review with the tutor
How you'll learn
A 100% online programme on the SEIUM campus, with downloadable material, recorded classes and an assigned technical tutor. There are no fixed timetables: students progress at their own pace within the intake, which makes it possible to combine studies with work.
Each module combines theory, guided reading of the original standard and applied exercises. The working software is NVIDIA Nsight Compute, Apache TVM, ONNX Runtime, NVIDIA TensorRT, PyTorch, in free, academic or evaluation versions. Case studies cover real compilation and deployment procedures for a detector on an accelerator, roofline analysis of a perception network, and a performance comparison between two platforms. Assessment is continuous, module by module, plus the final integrating project.
Build it end to end
Your final project follows the real engineering workflow, step by step, reviewed by your tutor.
Compare accelerator architectures for embedded AI.
Design neural networks adapted to hardware.
Use inference compilers and runtimes.
Apply a rigorous performance measurement methodology.
Profile and eliminate bottlenecks.
Where this takes you
Requirements
- Prior technical training or professional experience in the sector. A university degree is not required.
- Internet connection and a computer capable of running the programme's software.
- Sufficient Spanish to follow technical standards documentation.
Ready to start?
Send your application today. Our admissions team replies within 72 hours with your study plan and funding options.
Frequently asked questions
The programme comprises 800 hours in total, spread over 6 to 8 months. That is between 10 and 15 hours per week on average, compatible with a full-time job.
You will work with NVIDIA Nsight Compute, Apache TVM, ONNX Runtime, NVIDIA TensorRT, PyTorch. Free, academic or evaluation versions are used, and the tutor indicates the free alternative where one exists.
A SEIUM own diploma upon passing the continuous assessment and the integrating project. It is in-house training, not an official qualification from the Ministry.
Yes. Modules passed in a course or postgraduate diploma in the same area are recognised in the advanced programme. Check the recognition policy when you enrol.
Related programmes
Postgraduate Diploma in Precision GNSS/INS/Visual and RTK/PPP Fusion
Online 800-hour Postgraduate Diploma in Precision GNSS/INS/Visual Fusion and RTK/PPP. 8 modules with RTCM 10403.3, EN 16803-1:2020 and RTKLIB. Assigned tutor and open enrolment.
Postgraduate Diploma in Trajectory Planning (Sampling/Graph/Optimization)
Online 800-hour Postgraduate Diploma in Trajectory Planning (Sampling/Graph/Optimization). 8 modules covering ISO 26262:2018, ISO 21448:2022 (SOTIF) and OMPL. Assigned tutor and open enrolment.
Diploma in Operational Risk Management and Safety Metrics
Online postgraduate diploma of 800 hours in operational risk management and safety metrics. 8 modules covering risk in automated driving, hazard analysis, safety metrics and indicators,…
More in Autonomous, ADAS & Software-Defined Vehicle
All Autonomous, ADAS & Software-Defined Vehicle programmes → · How to apply · Fees & financing · Scholarships