AI Research Resident — In-Memory Inference Codesign

Normal Computing Corporation; New York (NY); On-site; USD 110,000 - 170,000; Full time

Job summary

Normal Computing is an applied AI company solving hard problems in AI and silicon. The AI Research Residency tracks the frontier of AI by joining a small cohort with a mentor, direct access to the team, and a clear arc from onboarding to publication.

The Algorithms Codesign track focuses on computational methods for efficient LLM inference on memory-based hardware, validating ideas in simulation, FPGA, and silicon as it arrives.

Qualifications

• Graduate degree (MS or PhD) in machine learning, applied mathematics, physics, electrical engineering, computer science, or related field.
• Strong Python skills; experience with a systems language (C++, Rust) or hardware-near algorithms.
• Experience turning research ideas into reproducible code and building publishable work.
• Familiarity with large-model inference concepts: attention, KV cache, long-context decoding.
• Publications are a plus but not required; solid research track record matters.

Responsibilities

• Develop inference algorithms for in-memory hardware; demonstrate first in simulation and then on hardware.
• Design numerical methods that fit the physics and exploit analog dynamics.
• Build evaluation suites and benchmarks to characterize algorithms on real hardware and in simulation.
• Translate workloads into hardware constraints for architecture teams.
• Prototype and iterate rapidly from simulation to FPGA to silicon.
• Drive a research question with rigorous experimental analysis and documentation.
• Co-author a paper and present at the residency colloquium.
• Present at the residency colloquium and engage with the broader Normal Computing research community.

Skills

Python

C++

Rust

Machine learning

Research

Education

MS or PhD in ML/CS/EE/Physics

Tools

Simulators

FPGA

Job description

Normal Computing is an applied AI company solving hard problems in AI and silicon. The AI Research Residency tracks the frontier of AI by joining a small cohort with a mentor, direct access to the team, and a clear arc from onboarding to publication.

The Algorithms Codesign track focuses on computational methods for efficient LLM inference on memory-based hardware, validating ideas in simulation, FPGA, and silicon as it arrives.