Job Title

About Company:Kinexcs is an AI-driven digital health platform and wearables company with a mission to enable and empower people for mobility and a better quality of life. It is focused on reducing the burden of musculoskeletal conditions, which affects about 25% of the world's population. Their 1st product, KIMIA Recover, a continuous monitoring device for the knee joint, has obtained CE marking and HSA approval and booked revenues from large implant companies and hospitals. It has recently won the Innomatch grant by the Temasek Foundation (one of the 6 among 256 companies across 32 countries), and is currently deployed in some of the largest public hospitals in Singapore. KIMIA Recover is a comprehensive recovery management solution that provides a reduction in the number of hospital visits and complications. The product was also the National Winner and International Top 20 of the prestigious James Dyson Award 2020 and has gained traction with major public hospitals and orthopaedic clinics. Their digital therapy platform, consisting of KIMIA Assess and KIMIA Move products, uses artificial intelligence technology for digital MSK assessment and therapy. This platform is capable of providing pre-consultation assessments and real-time exercise guidance and feedback to musculoskeletal patients and is being adopted by some of the largest private healthcare providers in the region, along with large corporate players.Job Summary:We are building a next-generation digital rehabilitation and fitness platform that combines computer vision, wearable sensors, and AI-driven analytics to improve patient recovery outcomes and long-term physical performance.As a CV/ML Scientist / Engineer, you will play a core role in transforming our current computer vision pipeline into a robust, clinically reliable, and scalable system, and in developing proprietary sensor + vision fusion algorithms that become a long-term IP moat for the company.This is not a “model fine-tuning only” role.You will work on real-world data, noisy environments, diverse patient populations, and production constraints in healthcare.Key ResponsibilitiesComputer Vision & Pose EstimationDesign, evaluate, and improve human pose estimation pipelines for rehab and fitness use casesEnhance robustness across:Clothing variations (loose clothing, hijabs, patterns)Lighting and camera placementDifferent body types, mobility limitations, and assistive devicesEvaluate, integrate and sutomize models such as:MoveNetHRNetMediaPipe (customized / extended)Other open-source or commercial CV frameworksSensor Fusion & Motion AnalysisDevelop sensor + computer vision fusion algorithms combining:IMU / wearable sensor dataRGB video-based pose estimationDesign temporal models to improve:Angle accuracy (ROM)Motion smoothnessRepetition countingExercise quality validationBuild algorithms resilient to partial occlusion or sensor dropoutRehabilitation & Fitness IntelligenceDevelop exercise validation and scoring algorithms:Range of motion (ROM)Speed, symmetry, control, compensation detectionContribute to personalized analytics using continuous ROM and exercise dataSupport wound image analytics and recovery progression insights (future roadmap)Data & ML EngineeringBuild pipelines for:Data preprocessing and labelingModel evaluation and benchmarkingOffline experimentation and online inferenceCollaborate with backend engineers to deploy models in:Real-time mobile pipelinesCloud-based batch and streaming workflowsBalance accuracy, latency, and compute costClinical & Product CollaborationWork closely with:Product managersUX designersClinicians and rehab specialistsTranslate clinical requirements into measurable technical metricsSupport validation studies and algorithm documentation (DHF-ready)Required QualificationsTechnical SkillsStrong background in Computer Vision and Machine LearningSolid experience with:PythonPyTorch or TensorFlowHands-on experience with:Human pose estimationTime-series analysisModel optimization for real-time inferenceUnderstanding of signal processing fundamentals (IMU / motion data)Practical ExperienceExperience deploying ML models into production, not just researchAbility to debug models using real-world data, not curated datasetsComfortable working with imperfect, noisy, and limited datasetsNice-to-Have (Strong Plus)Experience in healthcare, rehab, biomechanics, sports science, or fitness techSensor fusion experience (vision + IMU)Experience with:Temporal models (LSTM, TCN, transformers for motion)Edge or mobile ML optimizationFamiliarity with regulatory or compliance environments (HIPAA, GDPR, medical devices)Publications, patents, or open-source contributions in CV / ML