5,000+
combined beta users across both products
IntelliFit · AI Human Health Perception
AI that keeps understanding how people move, eat, and change
Visual AI, multimodal models, and wearable sensing turn real-world movement and nutrition into a personal health intelligence model.
Company Positioning
Not two isolated apps, but one system for understanding human behavior
IntelliFit is building a next-generation AI human health perception system. We enter frequent, verifiable movement and nutrition scenarios through CoachLink and FoodLink, then use wearables to capture continuous real-world signals.
Application layer: CoachLink understands training and coaching delivery; FoodLink understands eating behavior and nutrition feedback.
Perception layer: a compact AI fitness sensing module continuously captures movement, training, and body signals.
Intelligence layer: long-term behavior-state-outcome feedback creates a personal health model that changes with the user.
100K+
founder audience and organic launch channel
2 + 1
two app entries and one wearable direction
Application Layer
Two real products entering two high-frequency health behaviors
CoachLink and FoodLink are not the endpoint. They are application entries into a broader human behavior perception system, solving today’s problems while generating the real feedback required for future health intelligence.
3,500+ beta users
Movement intelligence
教链 · CoachLink
An AI movement training system for athletes, coaches, and studios. It connects planning, training logs, movement analysis, and coach operations so training can be understood, reviewed, and continuously improved.
- Training plans, execution logs, and cycle progress
- Movement video analysis and pose-estimation validation
- Coach operations, client collaboration, and delivery tools
Web and mini-program versions are live, with 3,500+ beta users validating the first entry into movement understanding.
Open CoachLink2,000+ beta users
Nutrition intelligence
食探 · FoodLink
An AI nutrition management system for individuals and organizations. It combines food recognition, nutrition analysis, and personalized guidance to understand each meal with less friction and connect eating behavior to long-term goals.
- Chinese meal recognition, portion estimation, and nutrition analysis
- Personal goals, health profiles, and continuous feedback
- Logging, analysis, community, and organizational scenarios
The website and mini-program are live, with 2,000+ beta users and a growing base of real meal images and correction feedback.
Visit FoodLinkPerception System
The next step in health management: from manual logging to continuous understanding
Traditional health apps wait for users to enter data. We want AI to sense behavior in the real world, understand physical state, and recommend the next executable action.
01 · Log
CoachLink and FoodLink capture training, meals, and subjective feedback as verifiable behavioral inputs.
02 · Sense
First-person vision, pose estimation, and multimodal sensors reduce logging friction and capture real behavior.
03 · Understand
Movement, nutrition, physical state, and context form a long-term behavior-state-outcome model.
04 · Decide
The system turns understanding into the next meal, workout, or adjustment aligned with the user’s goals.
Technology Roadmap
From application entries to wearable sensing and a human behavior intelligence platform
This is not a device bolted onto two apps. It is a staged path: validate users and scenarios, reduce sensing friction, then build a long-term model of human behavior across contexts.
Hardware concept · In validation
A compact AI fitness sensing module can clip onto workout clothing or equipment as an entry point for movement and training understanding. Its sensors and scenarios remain under validation.
01
Stage 1 · Application validation
CoachLink serves movement intelligence and FoodLink serves nutrition intelligence, validating real users, data, and feedback loops.
02
Stage 2 · Wearable sensing
Explore a clip-on AI fitness sensing module using vision, pose, and multimodal sensor fusion to understand training continuously.
03
Stage 3 · Behavior intelligence
Connect movement, nutrition, rehabilitation, and other signals into a personal health intelligence model that can expand to more human-understanding scenarios.
Product Evidence
Validate with real products before expanding the sensing boundary
Every screen below comes from a working CoachLink or FoodLink product. The application layer solves today’s problems and grounds the technology roadmap in real usage and feedback.
CoachLink · Training
Coach plans, training execution, and cycle progress
CoachLink · Toolbox
Video analysis, strength tools, and scenario-based utilities
CoachLink · Discovery
Coach discovery by direction before deeper conversation
食探 · Home
Meal logging, calorie budget, and micronutrient overview
食探 · Analysis
Health scores, AI analysis, and improvement priorities
食探 · Community
Public feed, check-in leaderboard, and food-sharing interactions
Founder And Team
A team at the intersection of AI, life science, and high-level training practice
IntelliFit starts from the founder’s AI and life-science research, serious strength-training practice, content reach, and the ability to build products with real users—not from an abstract health concept.
Validate real needs in apps before expanding wearable sensing.
Build behavior-state-outcome feedback before personal models.
Make guidance executable before asking the system to understand more.
Founder / AI & Product
Jianwen Ma
PhD student at Peking University’s Academy for Advanced Interdisciplinary Studies, with an automation background from Nanjing University of Aeronautics and Astronautics. He works across AI for Science and multimodal intelligence, while bringing serious strength-training practice, strong execution, and first-hand judgment on training methods, body management, and real user needs into product building.
Personal websiteWeb Full-Stack & Agent Delivery
Huang Zhelong (Jin hui)
Known as Jinhui, he writes technical articles and contributes to OpenMCP and slidev-ai, winner of Alibaba Best App Award. He provides web full-stack development plus frontier Agent R&D and iteration support. Hobbies: fitness, coffee, cycling, and coding.
Personal websiteBackend & Infrastructure / Performance
Ya Ning
Ya Ning focuses on reliable backend systems and AI infrastructure, covering backend deployment, cloud-native infrastructure, performance tuning, and reliability engineering. His work spans Go, Kubernetes, observability, and high-concurrency service practices to keep core product systems available and efficient.
Personal websiteCareers
Hiring early core members in Beijing
We are building an early AI + fitness and health company with real products, real users, early revenue, and active B2B opportunities. The bottleneck is not a lack of ideas; it is finding people who can own work and close loops.
In-person collaboration in Beijing matters. Product iteration, user feedback, content experiments, and commercialization all need high-frequency communication and people who can own a track over time.
AI perception and technical product
Build CoachLink and FoodLink while advancing movement video understanding, pose estimation, multimodal sensor fusion, wearable validation, and B2B delivery.
Computer vision, multimodal AI, frontend, full-stack, or hardware experience
Able to turn real scenarios into independent demos
GitHub, papers, competitions, hardware, or shipped work are strong pluses
Founder IP growth and commercialization
Build founder IP into a real commercial asset across content, user conversion, private community, consulting, training camps, live commerce, and partnerships.
Plan and review Douyin, Xiaohongshu, and Bilibili content
Design IP monetization and conversion paths
Revenue-share mechanisms can be discussed for clear commercial contribution
What we value most
Responsibility first Fast learning Attention to detail Stable commitment Timely communication Self-checking before delivery
Priority candidates
Based in Beijing and able to work together offline
Senior undergraduates or graduate students with most coursework completed
Real projects, shipped work, accounts, demos, or GitHub
Interest in AI, fitness, health, founder IP, commercialization, or early startups
Collaboration model
We usually start with a 3-7 day project-based trial so both sides can evaluate fit. High offline commitment can be paid as an internship; online or low-frequency work is better matched to workload, project delivery, and results. Strong long-term collaborators may enter core-member evaluation, with exceptional candidates considered for future partner roles.
How to apply
Please send your resume, project experience, portfolio, account links, demos, or GitHub to the personal email first. You can also add the work WeChat directly. Referrals of strong candidates are very welcome.
Collaboration
Help advance AI understanding of real human behavior
We welcome collaboration with coaches and sports organizations, enterprise and university health programs, computer-vision and wearable partners, and early teammates who want to build the system with us.
Movement and coaching
Training plans, movement analysis, coaching delivery, studios, and university sports.
Enterprise and university
Employee health, campus health, real-world pilots, and joint product validation.
Technology and hardware
Visual understanding, 3D pose estimation, multimodal sensors, and wearables.
Personal email
Company email
Work WeChat
HLG53589