Quantization & Numerics Engineer
Quantization & Numerics Engineer - AI Hardware / InferenceI'm working with an early-stage semiconductor company developing next-generation AI inference hardware, using novel transistor-level computing to improve efficiency and performance. They're looking for a Quantization & Numerics Engineer with experience in model optimisation, post-training quantization (PTQ) and hardware-aware numerical methods.
The role:
- Develop and own post-training quantization pipelines, including calibration, scaling and outlier handling.
- Design hardware-specific quantization techniques to optimise model efficiency while retaining accuracy.
- Build numerical emulators and evaluation frameworks to validate model performance.
- Collaborate with compiler, runtime and hardware engineers to optimise inference execution.
- What they're looking for:
- Hands-on experience in model quantization, low-precision inference and numerical optimisation.
- Experience quantizing multiple models and validating accuracy against full-precision baselines.
- Strong understanding of numerical precision, quantization errors and model-quality trade-offs.
- Experience evaluating LLMs across reasoning, coding and long-context tasks.
- Familiarity with hardware-aware quantization, numerical emulation or custom numeric formats is advantageous.
- Why consider it?
- Join an early-stage company developing fundamentally new AI inference hardware.
- Work directly with experienced chip architects and engineers.
- Own quantization techniques that directly influence next-generation AI circuit design.
- Interested? Apply directly or message me for a confidential discussion.