Research Engineer, Algorithms

About the role

Normal Computing is seeking a Research Engineer to develop computational methods that enable efficient AI inference on its thermodynamic ASIC hardware. You will work at the intersection of algorithm design and hardware co‑design, shaping how transformer and diffusion workloads run on stochastic analog memory.

Key responsibilities

• Develop algorithms for transformer inference on stochastic analog processing‑in‑memory hardware.
• Collaborate with hardware and architecture teams to influence chip capabilities.
• Design numerical methods that exploit thermal noise and analog dynamics.
• Build evaluation frameworks and benchmarks using real silicon or high‑fidelity simulation.
• Translate model workload insights into hardware design constraints and opportunities.
• Prototype rapidly as hardware evolves from simulation to silicon.

Required profile

• Deep understanding of large‑model inference, including attention mechanisms, KV cache, and long‑context decoding.
• Experience with inference optimisation techniques such as quantisation, sparsity, kernel fusion, and memory‑efficient attention.
• Familiarity with stochastic systems, probabilistic methods, numerical analysis, and analog computation.
• Proven ability to implement algorithms close to hardware, not just in high‑level frameworks.

Required skills

• Large‑model inference
• Attention mechanisms
• KV cache handling
• Long‑context decoding
• Quantisation
• Sparsity
• Kernel fusion
• Memory‑efficient attention
• Stochastic systems
• Probabilistic methods
• Numerical analysis
• Analog computation
• Hardware‑near algorithm implementation