Cephable

Job Summary:
Cephable is an innovative company focused on privacy-first, on-device AI solutions. They are seeking a Lead Machine Learning Engineer to advance their core ML systems, with responsibilities that include model development, optimization, and deployment for various applications.
Responsibilities:
• Design, train, fine-tune, and evaluate ML models for speech recognition, generative and reasoning models, and multimodal inference
• Adapt open-source and foundation models using Hugging Face and related tooling
• Translate research ideas into production-ready systems
• Optimize models for low-latency, low-power, offline execution
• Perform quantization, pruning, and distillation
• Deploy models via ONNX Runtime and OpenVINO targeting CPU, GPU, and NPU backends
• Build pipelines for training, evaluation, benchmarking, and regression testing
• Define and improve accuracy, latency, and resource metrics
• Partner with application and platform engineers to ensure seamless ML integration
• Communicate model performance, architectural decisions, and technical tradeoffs clearly to both technical and non-technical stakeholders
• Own Cephable’s ML architecture
• Set best practices and mentor team members
• Evaluate new tools, frameworks, and hardware
• Mentor engineers across the team on ML concepts and practices as the org grows
Qualifications:
Required:
• 4+ years of experience in machine learning or ML systems
• Strong PyTorch experience
• Hands-on experience with Hugging Face
• Production deployment using ONNX Runtime and/or OpenVINO
• Experience with acceleration frameworks like CUDA and GPU workflows
• Strong software engineering skills (Python, C++, or systems-level experience)
• Excellent communication skills — able to explain complex ML concepts, tradeoffs, and decisions clearly to engineers, product stakeholders, and non-technical partners alike
Preferred:
• Speech recognition or voice assistant experience
• LLMs, SLMs, or reasoning models
• Multimodal ML experience
• Edge or on-device AI background
• Experience with QNN, WinML, and CoreML
Company:
Cephable offers an ambient user interface platform that enables control of digital tools. Founded in 2023, the company is headquartered in Boston, USA, with a team of 11-50 employees. The company is currently Early Stage.