[SW] DNN Model Optimization Engineer

Frontier of ​On-device ​AI ​Semiconductors

for Everyone, ​Everywhere

About DeepX ​Co., ​Ltd.

DEEPX ​is a ​forward-thinking Series D startup ​architecting ​the infrastructure ​for the Physical ​AI era. ​

By ​delivering the ​world’s ​most ​energy-efficient NPU technology, ​we ​are solving the ​critical ​power ​and heat challenges ​of Generative ​AI ​to bring ​super-intelligence to ​every ​device, everywhere.

Our global ​leadership is ​validated by our record-breaking recognition as a multi-year CES Innovation Award honoree (2024 & 2026) and being named the 2024 Frost & Sullivan Company of the Year in the NPU sector. With an enterprise value approaching 1 trillion KRW, DEEPX offers a unique pre-IPO

opportunity to join a market leader defining the new industry standard for the $70B AI semiconductor market.

We are currently scaling toward mass production of our flagship DX-M1 (Samsung 5nm) with over 50 global projects scheduled, while engineering next-generation 2nm solutions to support 100B parameter Large Language Models (LLM) at the edge. ☞ Link

★ If you want to be part of world-class innovation? Please talk with us.

★Explore our journey: The DEEPX White Paper ☞ Link

Role Overview

The Software Group at DEEPX plays a critical role in realizing DEEPX's mission to lead the world in core AI processor technology. The team develops essential software technologies that allow customers building AI-powered applications to easily and efficiently integrate DEEPX’s advanced processors into their products.

From firmware and drivers to compilers (DX COM), runtime systems (DX RT), and code design for DNN model training frameworks, the group is responsible for building the full software stack needed to deploy DEEPX NPUs into real-world products.

The DNN Model Team focuses on converting SOTA (state-of-the-art) and customer-developed DNN models into DEEPX’s proprietary intermediate representation. The team also develops optimizations at the model level to ensure the best possible performance on DEEPX NPUs.

We are looking for DNN Model Engineers who have a deep understanding of AI computation and are passionate about model optimization and hardware-aware performance tuning. You will analyze and optimize customer models and contribute to the front-end compiler infrastructure. This is a unique opportunity to work on cutting-edge AI models and help implement them on innovative high-performance, low-power AI platforms.

Responsibilities

□ Analyze DNN model operations and develop optimization/quantization techniques

□ Research and apply GPU-level optimization methods, including tracking the latest academic and industry trends

□ Study and develop optimization methods for modern DNN models

□ Analyze various DNN compilers and stay updated with related literature

□ Perform algorithm-level optimizations for DEEPX software code

Qualifications

□ ​Experience in training and optimizing deep learning models using frameworks such as PyTorch or TensorFlow

□ Hands-on experience with recent DNN models (e.g., object detection, LLMs, diffusion models)

□ Experience with at least one model compression technique such as quantization, pruning, or knowledge distillation (KD)

Preferred Qualifications

□ Contributions to open source software

□ Experience optimizing and deploying DNN models across various hardware platforms (CPU, GPU, FPGA)

□ Experience using open-source toolkits such as OpenVINO, AIMET

□ Experience optimizing GPU kernels using CUDA, Triton, etc.

Recruitment Process

□ Application Review - (Phone Interview) - Technical Interview - Organizational Culture Fit Interview - CEO Interview - Reference Check / Compensation Discussion

※ The recruitment process may vary depending on the position and application content.

※ Candidates with less than 3 years of experience are required to submit their academic transcripts.

Employment Type

□ Full-time (3-month probationary period with 100% compensation)

Working Hours

□ Monday to Friday, 9:00 AM – 6:00 PM (Lunch break: 12:00 PM – 1:00 PM)

Notes

□ If any false information is found in the application or onboarding documents, the job offer may be withdrawn even after confirmation.

□ A 3-month probationary period applies after joining, with no reduction in salary or benefits.

Benefits

□ 모든 정규직 입사자에게 연봉 수준의 스톡옵션 부여

□ 최신 사양 장비 및 최고의 근무 환경 제공 (최신 노트북, 높낮이 조절식 스탠딩 데스크, 모니터암, 듀얼모니터 등 제공)

□ 점심식사 + 아침 & 저녁식사 지원

□ 스낵, 아이스크림, 음료 등 사내 카페 무제한 간식 제공

□ 사우나가 포함된 피트니스 비용 지원

□ 연 1회 종합건강검진 지원 (배우자 포함)

□ 생일, 결혼기념일, 크리스마스이브 축하금 지급 및 조기퇴근 제공

□ 설/추석 명절 상여금 지급

□ 축하와 위로를 위한 경조휴가 및 경조금 지원