Small Language Model (SLM) Developer – On-Device AI

Location: Remote – US, Americas, and Europe preferred
Type: Full-time

Compensation: Base salary: $150,000 – $200,000 per year.

Sensory is looking for a Small Language Model Developer to build and optimize compact, high-performance language models that run directly on devices and integrate tightly with Sensory STT and Sensory’s wake word and biometric stack. You will help define the future of efficient, private, and reliable conversational AI at the edge.

Responsibilities

• Design, train, fine-tune, and evaluate small language models optimized for on-device deployment across multiple platforms and languages.​
• Work closely with the STT and wake word teams to create seamless pipelines for transcription, NLU, and response generation without hallucinations.
• Implement techniques such as quantization, pruning, distillation, and custom architectures to reduce model size while preserving accuracy.​
• Develop domain-specific micro-NLU and SLM components for key use cases (e.g., automotive, smart home, hearables, enterprise devices).
• Build robust evaluation suites, benchmarks, and tooling to measure performance under real-world noise, accent, and latency constraints.
• Collaborate with product and customer teams to translate requirements into model specs and deployment strategies.

Requirements

• 3+ years of experience in NLP, machine learning, or applied deep learning, ideally with a focus on model efficiency.
• Strong background in transformer-based architectures and modern techniques for compression and optimization of language models.
• Proficiency in Python, C/C+ and deep learning frameworks such as PyTorch or TensorFlow.
• Familiarity with deploying models on mobile, embedded, or low-power platforms, including an understanding of memory and compute budgets.
• Experience working with multilingual data and evaluation across many languages is a plus.​
• Excellent command of conversational and relevant technical English.

Nice to Have

• Prior work with speech-to-text + NLU stacks and understanding of how external STT can improve LLM/SLM performance.​
• Contributions to open-source projects, research publications, or demonstrable side projects in SLMs, edge AI, or speech/voice models.