Please click the "Apply" button above to submit your resume. To complete your application, you must also sign the Personal Information Collection and Use Consent form here. Your application will be finalized only after the signature is submitted.

This job posting is for Softionics, and is supported by KOTRA Silicon Valley.

본 공고는 Softionics의 요청으로 코트라 실리콘밸리가 지원합니다.

[About Company]

Softionics builds a vision-free near-field sensing layer for Physical AI, capturing space, time, and physical interaction data that vision-systems cannot. Our technology enables robots and smart devices to sense proximity, motion, and contact dynamics in real time. Softionics is a deep-tech startup developing non-contact sensing solutions based on proprietary electric field sensor technology, targeting robotics, wearables, and human-machine interaction applications. We are advancing from sensor circuit prototyping to full custom IC design.

[Job Responsibilities]

We are seeking an ML & Data Research Engineer to develop and advance AI models that leverage our unique electric field sensor modality. A key research direction is extending Vision-Language-Action (VLA) models — such as OpenVLA, RT-2, and π0 — by incorporating electric field sensor data as an additional input channel (VLAE: VLA + Electric field). This is a strong research differentiator: current VLA models rely on camera alone and struggle with near-contact perception for dexterous manipulation, which is exactly where our sensor excels.

• Design algorithms for electric field sensor data (signal processing, feature engineering, inverse problem solving)

• Develop and train ML models for multimodal data (electric field sensor + camera)

• Research and implement models for hand motion tracking, object localization, and shape estimation

• Fine-tune open-source VLA models (OpenVLA, Octo, π0, etc.) and extend them with electric field sensor modality

• Develop proximity-aware grasp policies for robotic hands using sensor input

• Pre-train models on synthetic datasets generated by BEM simulation

• Design experiments (ablation studies, hyperparameter tuning, quantitative evaluation)

• Collaborate with the Systems Engineer (Position 2A) on dataset requirements and quality

• Publish research results (internal reports, academic papers, patents)

[Location] Bldg 35 Room 524-3, 1 Gwanak-ro, Gwanak-gu, Seoul, Republic of Korea, 08826

[Job Type & Pay Range]

• Conversion Internship

• Negotiable based on experience and skill set

[Required Qualifications] 

• ML/DL model design and implementation (PyTorch or TensorFlow)

• Experience developing models on time-series sensor data or physical signal data

• Ability to design experiments and analyze results quantitatively

• Python data analysis and visualization (NumPy, Pandas, Matplotlib, etc.)

• Reproducible experiment management with Git

[Preferred Qualifications] 

• Experience with VLA (Vision-Language-Action) models: RT-2, OpenVLA, π0, Octo, or similar

• Multimodal foundation model fine-tuning experience (especially adding novel sensor modalities)

• Robot Learning, Imitation Learning, or Reinforcement Learning experience

• Physics-Informed Neural Networks or Inverse Problem research experience

• Electric field, capacitive, impedance, or tactile sensor data analysis experience

• Computer vision experience: pose estimation, hand tracking, point cloud processing

• BEM/FEM simulation data utilization experience

• Publications at CVPR, NeurIPS, ICRA, IROS, CoRL, or equivalent

• M.S. or Ph.D. in EE, CS, AI, or Robotics preferred

[Benefits]

• Competitive salary commensurate with experience and skills

• Stock options (based on contribution and negotiation, Not Right Now)

• Flexible working hours (autonomous schedule)

• Generous paid leave policy with tenure-based additional leave

• Snacks and overtime meal allowance provided

• Full support for job-related books, online courses, and conference attendance

• AI development tools provided (Claude Code, Copilot, etc.)

• Opportunities to participate in domestic and international R&D collaboration projects

• Opportunity to work on cutting-edge electric field sensor technology at a deep-tech startup

• Direct exposure to the full product development cycle from research to commercialization