NPU Software Engineer - Kernel

New roles land in Following.

Rebellions | KoreaPosted 7 months ago

Apply on Rebellions's career page

Market pay check

Matched to Computer systems and software professionals

Source: Ministry of Employment and Labor wage information system (wage.go.kr), wage distribution by occupation.

What you'd do

Design and develop compute libraries (BLAS,DNN) optimized for proprietary instruction set architecture.
Expand functional coverage of operations by addressing tensor shapes,precision,numerical stability.
Optimize operation-level performance based on heterogeneous compute resource utilization and workload characteristics.

What they want

Master's degree or higher in Electrical Engineering,Computer Science,or related field.
In-depth understanding of computer architecture and parallel programming.
Strong understanding of neural network operations,high-level concepts,low-level computational dataflows.

Nice to have

Broad knowledge of deep learning models across computer vision,NLP,speech recognition.
Experience developing high-performance custom kernels for generative AI on GPUs or NPUs.
Experience with model and layer-level optimization: sparsity,reduced precision,layer decomposition.

Seoulstart's read

Visa tier (likely): E-7 specialty / professional (likely)
Korean language: Korean not required
English friendliness: High: JD in English, foreign-friendly signals
Context: Rebellions is a Seoul-based AI chip startup; English-primary technical environment typical for deep-tech companies,but E-7 visa sponsorship standard for Master's-level engineering hires at domestic tech firms.

This is Seoulstart's analysis of the public job description, not the employer's stated policy. Verify visa sponsorship, language requirements, and remote allowances directly with the recruiter.

Original posting from Rebellions

Responsibilities and Opportunities

Design and develop compute libraries (e.g., BLAS, DNN libraries) consisting of neural network operations optimized for Rebellions’ proprietary instruction set architecture (ISA)
Expand the functional coverage of each operation by addressing operation-specific constraints, including tensor shape variations, precision management, and numerical stability
Improve utilization of heterogeneous compute resources by optimizing operation-level performance based on workload characteristics
Design and implement architecture-aware custom kernels based on a deep understanding of dynamic execution patterns in generative AI workloads
Contribute to hardware–software co-design by identifying and proposing critical features for next-generation compute architectures

Key Qualifications

Master’s degree or higher in Electrical Engineering, Computer Science, or a related field
In-depth understanding of computer architecture and parallel programming
Strong understanding of neural network operations, including both high-level concepts and low-level computational dataflows
Strong analytical, troubleshooting, and performance optimization skills
Proficiency in C++ and Python

Ideal Qualifications

Broad knowledge of deep learning models across multiple domains, including computer vision, natural language processing, and speech recognition
Experience in developing high-performance custom kernels for generative AI workloads on GPUs or NPUs
Experience in model- and layer-level optimization techniques for computational efficiency, such as sparsity, reduced precision, and layer decomposition
Experience with architecture-specific parallel programming and hardware acceleration frameworks, including SSE/AVX (x86), NEON (AArch), and CUDA/OpenCL (GPU)
Academic background in computer architecture is preferred

전형절차

서류전형 > On-line 인터뷰 > On-site 인터뷰(과제 포함) > Culture-fit 인터뷰 > 처우 협의 > 최종 합격
전형절차는 직무별로 다르게 운영될 수 있으며, 일정 및 상황에 따라 변동될 수 있습니다.
전형 일정 및 결과는 지원 시 작성하신 이메일로 개별 안내드립니다.

참고사항

본 공고는 모집 완료 시 조기 마감될 수 있습니다.
지원서 내용 중 허위사실이 있는 경우에는 합격이 취소될 수 있습니다.
채용 및 업무 수행과 관련하여 요구되는 법령 상 자격이 갖추어지지 않은 경우 채용이 제한될 수 있습니다.
보훈 대상자 및 장애인 여부는 채용 과정에서 어떠한 불이익도 미치지 않습니다.
담당 업무 범위는 후보자의 전반적인 경력과 경험 등 제반사정을 고려하여 변경될 수 있습니다. 이러한 변경이 필요할 경우, 최종 합격 통지 전 적절한 시기에 후보자와 커뮤니케이션 될 예정입니다.
채용 관련 문의사항은 아래 메일 주소로 문의바랍니다.
recruit@rebellions.ai

Open posting

Related Seoulstart guides

Similar roles on Seoulstart

Browse all English-friendly jobs in Korea

What you'd do

Design and develop compute libraries (BLAS,DNN) optimized for proprietary instruction set architecture.

Expand functional coverage of operations by addressing tensor shapes,precision,numerical stability.

Optimize operation-level performance based on heterogeneous compute resource utilization and workload characteristics.

Seoulstart's read

Visa tier (likely)

E-7 specialty / professional (likely)

Korean language

Korean not required

English friendliness

High: JD in English, foreign-friendly signals

Context

Rebellions is a Seoul-based AI chip startup; English-primary technical environment typical for deep-tech companies,but E-7 visa sponsorship standard for Master's-level engineering hires at domestic tech firms.

This is Seoulstart's analysis of the public job description, not the employer's stated policy. Verify visa sponsorship, language requirements, and remote allowances directly with the recruiter.