Remote AI Engineer (Deep Learning Algorithm/ Python)

Top 3 reasons to join us

• Build Real AI Products That Reach Users:
• High Ownership, High Impact
• Startup Speed with Global Vision

Job description

VisionTech is building an algorithm team capable of truly integrating deep learning into products and onto terminal devices. This role is not just about connecting models or tuning packages; it requires continuously improving model performance under the constraints of terminal computing power, memory, latency, and product limitations. We are looking for someone who possesses the complete capability from problem definition and method design to experimental verification. We hope you don't just use models, but deeply understand the principles behind them, can independently dismantle abstract problems, propose hypotheses, design experiments, and land research results into real products.

Responsibilities:

• Challenges: Continuously improve model performance under limited computing power and memory, advancing from demos to mass-producible, maintainable, and continuously evolving products, and flexibly responding to product issues reported by customers.
• Edge AI Design: Responsible for the design, training, optimization, and deployment of Edge AI models to improve product performance under terminal platform power constraints.
• Systemic Improvement: Systematically improve model performance based on feedback and demo requirements through model architecture, training methods, loss design, data preprocessing, data cleaning, and augmentation strategies.
• Performance Balancing: Analyze performance bottlenecks and balance accuracy, latency, memory, and power consumption according to product requirements and hardware constraints.
• Technical Research: Independently read papers, track technical progress, and evaluate applicability and cost-effectiveness for actual products. Deconstruct vague or abstract product problems into verifiable technical questions, proposing improvement directions and planning experiments.

Your skills and experience

• Location: Not based in Taiwan. We are looking for partners permanently based overseas (i.e., not in Taiwan), regardless of nationality.
• Deep Learning Fundamentals: Familiarity with basic principles including, but not limited to:
  • Neural network optimization
  • Regularization
  • Generalization
  • Model evaluation and error analysis
• Development Experience: Practical experience in model development with the ability to independently complete:
  • Problem definition
  • Data analysis
  • Model training
  • Experimental design
  • Result interpretation
• Frameworks: Familiarity with at least one mainstream deep learning framework, such as PyTorch.
• Engineering Skills: Solid programming skills and engineering habits, capable of writing maintainable, reproducible, and scalable research and training code.
• Analytical Ability: Ability to read technical papers and transform them into executable experimental plans. Possess independent thinking and problem-deconstruction skills rather than relying solely on existing recipes or ready-made models.
• Edge Constraints: Understanding of common constraints and challenges when landing models on edge/embedded platforms, such as:
  • Latency
  • Model size
  • Memory footprint
  • FLOPs

Preferred Qualifications:

• Publications/Patents: Previously published patents or deep learning papers in international conferences (e.g., NeurIPS, ICLR, CVPR, etc.).
• Specialized Experience:
  • Model compression, pruning, and quantization
  • Knowledge distillation
  • On-device inference optimization
  • Research/development experience in generative models
• Deployment: Experience deploying models to actual terminal platforms.
• Data Strategy: Familiarity with the impact of data preprocessing and quality on model performance; experience in data engineering or data strategy optimization.
• Project Lifecycle: Experience advancing research topics or production projects from 0 to 1.
• Resource Management: Successfully achieved acceptable model performance and deployment costs in resource-constrained scenarios.

Ideal Candidate Attributes:

• Not satisfied with just "getting the model running," but asks "why it works, why it failed, and how to make it better."
• Actively establishes baselines, proposes hypotheses, and designs verification processes when facing abstract problems.
• Finds balance between research depth and product reality rather than just chasing paper ideas or simple engineering stacking.
• Passionate about deep learning and willing to translate understanding into real-world systems.

Qualifications & Package:

• Education: Master's degree or above in Computer Science, Electrical Engineering, Data Science, or related fields.
• Location: Not based in Taiwan.
• Compensation: Cash compensation will be based on your strength and location. We provide a "competitive" equity package (out company believes that those with IPO experience in this industry understand the high potential of stock options).

Why you'll love working here

• Build Real AI Products That Reach Users: Your work contributes directly to real-world AI imaging technologies integrated into consumer devices and applications.
• High Ownership, High Impact: Every team member plays a critical role in product direction, technical innovation, and strategic decision-making.
• Startup Speed with Global Vision: Experience the fast-paced energy of a startup while working on technologies designed for international markets and large-scale commercialization.
• Work on cutting-edge AI imaging technologies that directly shape future smartphone experiences
• Opportunity to participate in end-to-end product innovation, from algorithm concepts to commercialization
• Small but highly agile international team with strong ownership culture and fast execution
• Exposure to collaborations with smartphone ecosystem partners and global technology companies
• Flexible and innovation-driven environment where ideas can rapidly become real products
• Opportunity to contribute to IP-driven products with strong market differentiation potential
• Cross-functional collaboration across AI, product design, engineering, and business strategy