About Zepp Health and Amazfit

At Zepp Health, innovation meets wellness to redefine what's possible in health, fitness, and wellness through our groundbreaking Amazfit smartwatches and wearable technologies. Since our inception in 2013, we've been at the forefront of merging cutting-edge science and technology to develop wearables that not only stand out for their style but also their performance. Amazfit, our global consumer brand, is more than just a product line; it's a commitment to empowering individuals to elevate their game, offering an array of devices from smartwatches to earbuds and fitness gear that seamlessly integrate into the fabric of daily life.

The Zepp Cloud and Big Data team already has strong backend and data engineering capabilities. The team has also delivered products built on commercial foundation models and has practical experience in dataset development, offline evaluation, and A/B testing. However, we currently lack senior talent with a model training background, which makes it difficult to evaluate the evolution and boundaries of different technologies from a holistic perspective.

The purpose of this hire is therefore not to add another engineer who knows how to call an LLM API, but to bring in a leader who can take unified ownership of AI product direction, technical strategy, quality standards, and team capability development.

Role Positioning. We need a hands-on Applied AI leader—a player-coach who can personally contribute to critical prototypes, architecture design, and technical reviews. This person should combine deep ML and model training expertise with a strong focus on production delivery, user value, and business outcomes.

Why a Strong ML Background Matters. The team needs someone who can determine whether a performance issue originates from task definition, data quality, retrieval, context, tool orchestration, product interaction, foundation-model capability, or the training approach. This person must be able to judge whether fine-tuning, distillation, an open-weight model, or a domain-specific model will deliver measurable incremental value, while also covering Zepp’s long-term opportunities in time-series data, sensor data, multimodal systems, personalization, and on-device models. The value of model training experience is not to push the team toward building models by default.

It is to help the team decide when not to train—and, when training is justified, how to do it effectively.

Core Selection Principle. We will prioritize candidates who have owned the complete AI/ML product lifecycle: starting from the user problem and success criteria, and continuing through data strategy, technical selection, evaluation, launch, monitoring, regression testing, and ongoing iteration. The right candidate can make evidence-based build, buy, or partner decisions across commercial foundation models, RAG, agents, conventional ML, model fine-tuning, open-weight models, and domain-specific models. They can also connect offline quality metrics to online user outcomes, business impact, and total cost of ownership.

Key Responsibilities

Define the Applied AI Strategy

- Develop and own the technical strategy for AI-powered health, fitness, coaching, and conversational experiences.




- Translate product and business opportunities into a prioritized AI roadmap with clear success metrics.
- Identify where AI can create differentiated user value and where deterministic software, rules, or conventional machine learning are more appropriate.
- Make evidence-based build, buy, or partner decisions across commercial models, open-weight models, fine-tuning, distillation, and domain-specific model development.
- Evaluate technical options based on product quality, differentiation, latency, privacy, safety, operational complexity, and total cost of ownership.

Bring AI Capabilities into Production
- Lead the architecture and delivery of production AI systems using foundation models, retrieval, tool use, structured outputs, agent workflows, and other appropriate technologies.
- Guide the evolution of existing NLU and conversational systems, including intent, entity, dialogue-state, and response-generation components.
- Design systems that combine LLM capabilities with trusted data sources, deterministic business logic, domain models, and user context.
- Establish strategies for model routing, fallback, caching, graceful degradation, and provider resilience.
- Ensure AI capabilities integrate effectively across cloud services, mobile applications, wearable devices, and, where appropriate, on-device models.

Establish Evaluation and Data Standards
- Define dataset, annotation, evaluation, and release standards for AI features.
- Build evaluation frameworks covering success cases, failure cases, boundary conditions, adversarial inputs, and safety-critical scenarios.
- Combine automated evaluation, human expert review, user feedback, and online experimentation.
- Connect offline model and system metrics to A/B tests, user outcomes, and business impact.
- Establish reliable feedback loops that turn production failures and user feedback into validated product and system improvements.
- Ensure training, validation, and test data are properly governed and protected against contamination or leakage.

Own Production Quality

- Define quality gates for models, prompts, retrieval systems, tools, and end-to-end AI experiences.
- Own production targets for quality, availability, latency, throughput, and cost.
- Establish observability for model behavior, retrieval quality, tool execution, safety failures, and operational incidents.
- Ensure model, prompt, dataset, knowledge-base, and tool versions are traceable and reproducible.
- Define deployment, rollback, regression-testing, and incident-response practices for AI systems.
- Monitor model and provider changes and prevent unvalidated behavior changes from reaching users.

Define Health and Safety Boundaries
- Establish explicit boundaries for AI-generated health, fitness, recovery, sleep, and coaching experiences.
- Define how systems handle uncertainty, missing or conflicting sensor data, and low-confidence results.




- Ensure high-risk scenarios are appropriately constrained, escalated, reviewed, or declined.
- Work with medical, legal, privacy, security, and regulatory stakeholders to define launch requirements.
- Ensure sensitive user data is handled appropriately across internal systems and external model providers.
- Apply responsible AI principles without losing sight of usability and product value.

Lead and Grow the Team
- Lead a team of ML and AI engineers through hands-on technical direction, design reviews, mentoring, and example.
- Develop existing backend and data engineers who are moving into applied AI work.
- Identify capability gaps and help define the future structure of the AI organization.
- Create a culture of rigorous experimentation, clear evidence, practical engineering, and shared accountability.
- Communicate technical opportunities, limitations, risks, and investment decisions clearly to both engineering and executive stakeholders.
- Collaborate effectively across global teams and time zones.

Qualifications and Success Measures

Basic Qualifications

- Typically 8+ years of professional experience in software engineering, machine learning, data science, or AI systems, including substantial experience delivering production AI or ML products.
- A bachelor’s degree in computer science, machine learning, engineering, or a related field, or equivalent practical experience.
- Proven ownership of at least one complete AI or ML product lifecycle, from problem definition and data strategy through evaluation, production deployment, monitoring, and iteration.
- Strong understanding of machine learning and deep learning fundamentals, with hands-on experience training, adapting, evaluating, and deploying models.
- Production experience with foundation-model applications, including several of the following: retrieval-augmented generation, tool and function calling, agent workflows, structured generation, context and memory management, model routing, fine-tuning or distillation, and open-weight model deployment.
- Demonstrated ability to determine whether a system failure originates from data quality, model capability, retrieval, orchestration, product design, or software implementation.
- Experience defining task-specific evaluations and connecting offline quality metrics to online product experiments.
- Experience making technical decisions involving quality, latency, cost, privacy, reliability, and operational complexity.
- Experience leading senior engineers and influencing cross-functional product roadmaps.
- Strong ownership, analytical thinking, communication, and execution skills.
- Ability to remain hands-on while providing technical and organizational leadership.

Benefits of Working At Zepp Health:

- Competitive salary, Vacation day, sick day and a remote-friendly culture
- Health insurance, Vision insurance, Dental insurance,
- Year-end Bonus pay
- Other Perks

Zepp Health is an Equal Opportunity employer and welcomes everyone to our team. If you need reasonable accommodation at any point in the application or interview process, please let us know. In your application, please feel free to note which pronouns you use (for example: she/her/hers, he/him/his, they/them/theirs, etc).

📌 Head of Applied AI Engineer (British Columbia)
🏢 Zepp Health
📍 British Columbia