Hands-on experience around 5 years with PyTorch and PyTorch Lightning for building, training, and extending production pipelines, along with Python proficiencyНавичка · Обов’язково
Strong applied deep learning background grounded in numerical/signal time-series data (IMU, EMG, audio, sensor fusion, biosignals, or industrial time-series) using architectures such as 1D-CNNs, RNNs/GRUs, dilated convolutions, or Temporal Transformers.Інше · Обов’язково
Comfortable with ensembling and cascade-style model architectures — combining multiple models' outputs and evaluating latency/complexity vs. performance trade-offs.Інше · Обов’язково
Solid grasp of evaluation metrics for imbalanced, multi-class time-series classification/segmentation across hundreds of classes (precision/recall trade-offs, label noise, sensor placement variance).Інше · Обов’язково
Able to work independently with a large, configuration-driven codebase, respecting established architectural conventions and software design patterns.Інше · Обов’язково
Strong Upper-Intermediate (B2+) or higher — comfortable with daily written and spoken technical communication.Інше · Обов’язково
the schedule can be adapted.Інше · Обов’язково
Good to knowІнше · Обов’язково
Experience with AWS Batch, Step Functions, S3, and Docker containerization for scalable model trainingНавичка · Обов’язково
On-device/edge inference: CoreML, TFLite, ONNX Runtime Mobile, or similar export/runtime experienceІнше · Обов’язково
Model compression: quantization, pruning, distillationІнше · Обов’язково
Wearables or human activity recognition (HAR) experience specificallyІнше · Обов’язково
Weights & Biases or comparable experiment-tracking disciplineІнше · Обов’язково
Mobile app-side familiarity (Swift/Kotlin)Навичка · Обов’язково
Working hours: 15:00 – 23:00 (Kyiv time) — preferred working hoursІнше · Буде плюсом
We are looking for a Movement Detection ML Engineer to drive the research and development of exercise-recognition models from IMU sensor data (accelerometer/gyroscope) and scale coverage from our current ensemble to several hundred exercises, while taking inference live on-device.
Key responsibilities
Increase accuracy and expand coverage to several hundred exercises. You will fine-tune existing models, redesign ensembling strategies, evaluate architectures suited for long-tail, imbalanced multi-class time-series classification, and adapt training regimes (data augmentation, sampling, curriculum learning) for newly annotated data.
Port the server-side inference pipeline to run live on iOS/Android devices under strict latency, memory, and compute constraints, handling real-world sensor noise and dropped samples through model export, quantization, and distillation.
Ensure all architectural and ensembling wins on offline accuracy respect mobile edge-budget constraints right from the design phase.
Requirements
Hands-on experience around 5 years with PyTorch and PyTorch Lightning for building, training, and extending production pipelines, along with Python proficiency
Strong applied deep learning background grounded in numerical/signal time-series data (IMU, EMG, audio, sensor fusion, biosignals, or industrial time-series) using architectures such as 1D-CNNs, RNNs/GRUs, dilated convolutions, or Temporal Transformers.
Comfortable with ensembling and cascade-style model architectures — combining multiple models' outputs and evaluating latency/complexity vs. performance trade-offs.
Solid grasp of evaluation metrics for imbalanced, multi-class time-series classification/segmentation across hundreds of classes (precision/recall trade-offs, label noise, sensor placement variance).
Able to work independently with a large, configuration-driven codebase, respecting established architectural conventions and software design patterns.
Strong Upper-Intermediate (B2+) or higher — comfortable with daily written and spoken technical communication.
Working hours: 15:00 – 23:00 (Kyiv time) — preferred working hours, but the schedule can be adapted.
Good to know
Experience with AWS Batch, Step Functions, S3, and Docker containerization for scalable model training
On-device/edge inference: CoreML, TFLite, ONNX Runtime Mobile, or similar export/runtime experience
Model compression: quantization, pruning, distillation
Wearables or human activity recognition (HAR) experience specifically
Weights & Biases or comparable experiment-tracking discipline
Mobile app-side familiarity (Swift/Kotlin)
About the company Kevych Solutions
We are a software development company that provides digital technology services. From mobile and web development to custom software solutions. We're dedicated to delivering innovative, high-quality solutions tailored to your unique needs.