AI Research Residency: In-Memory Codesign for LLMs

Normal Computing Corp.; New York (NY); Hybrid; USD 90,000 - 130,000; Full time

Job summary

Normal Computing invites exceptional researchers to apply to its AI Research Residency. This hands-on program pairs residents with mentors to tackle frontier problems in inference algorithms and hardware co-design, with a clear path to publication and a program colloquium.

The Algorithms Codesign track focuses on rethinking how attention, memory access, and long-context decoding run inside memory, validated in simulation, FPGA, and eventually silicon.

Qualifications

• Graduate degree (MS/PhD or equivalent) in ML, math, physics, EE, CS or related field is expected.
• Strong research track record; publications are a plus.
• Solid Python skills; systems or hardware-oriented algorithm experience is valued.

Responsibilities

• Develop and validate inference algorithms for in-memory compute on AI hardware.
• Design and analyze numerical methods that suit hardware constraints and analog dynamics.
• Build and contribute to evaluation suites and benchmarks for real hardware and simulators.
• Translate workloads into hardware constraints to guide architecture teams.
• Prototype and iterate from simulation to FPGA and silicon.

Skills

Python

Research experience

Publications beneficial

Education

MS or PhD (or equivalent) in ML/CS/related

Tools

C++

Rust

Job description

Normal Computing invites exceptional researchers to apply to its AI Research Residency. This hands-on program pairs residents with mentors to tackle frontier problems in inference algorithms and hardware co-design, with a clear path to publication and a program colloquium.

The Algorithms Codesign track focuses on rethinking how attention, memory access, and long-context decoding run inside memory, validated in simulation, FPGA, and eventually silicon.