AI Research Resident (Algorithms Codesign) at NormalComputing · LumiJobs

Responsibilities

Develop inference algorithms for in-memory hardware. Understand which parts of LLM workloads are best suited to in-memory execution and develop algorithms that exploit it natively, demonstrated first in simulation and then on hardware.
Design numerical methods that fit the physics. Design methods that exploit thermal noise and analog dynamics rather than working around them.
Build evaluation suites and benchmarks. Help build the evaluation frameworks and benchmarks that characterize how algorithms behave on real hardware and in simulation, and tell us honestly whether any of it actually works.
Translate workloads into hardware constraints. Turn insights about model workloads into constraints and opportunities for the hardware and architecture teams.
Prototype and iterate rapidly. Move with the hardware as it evolves from simulation to FPGA to silicon.
Drive a research question of your own. Partner with researchers and engineers to scope, run, and iterate on an original technical investigation, with rigorous experimental analysis and documentation along the way.
Co-author a paper. Work with the team to write up your findings for submission to a relevant venue, with mentorship on framing, experiments, and technical writing. One of the two milestones every resident builds toward.
Present at the residency colloquium. Share your work and thinking with the broader Normal Computing research community at the program's capstone event, and get real-time feedback from people building this technology every day.
What Would Make You a Great Fit
Currently pursuing or recently completed a graduate degree (MS or PhD, or equivalent research experience) in machine learning, applied mathematics, physics, electrical engineering, computer science, or a related field. Publications are a strong plus, but not required at the resident level; a solid research track record is what matters.
A working understanding of large model inference: attention mechanisms, KV cache, long-context decoding, memory bandwidth constraints. Deep production experience isn't expected at the resident level, but the intuition should feel familiar.
Some exposure to inference optimization (quantization, sparsity, kernel fusion, or memory-efficient attention), or to stochastic systems, probabilistic methods, numerical analysis, or analog computation.

AI Research Resident (Algorithms Codesign)

Responsibilities

Company

NormalComputing

More roles at NormalComputing