AI/ML Engineer

Cadence ABA builds AI-native clinical software for ABA therapy, capturing therapy sessions by voice, structuring the resulting trial data, tracking mastery against clinical targets, and drafting billable session notes with CPT codes mapped to the work performed. We're looking for an AI/ML Engineer with a focus on speech to help build and improve the ambient voice capture and transcription systems that sit at the core of the product.

What you'll do

• Build and improve speech-to-text and audio understanding pipelines that transcribe and resolve clinical events from natural, in-session speech
• Improve model robustness to real-world clinical environments: overlapping speech, background noise, varied accents, and offline or low-connectivity settings
• Design and run evaluation frameworks to measure transcription accuracy, latency, and downstream impact on clinical data quality
• Collaborate with clinical and product teams to translate ABA-specific terminology and session structure into model training data and evaluation sets
• Work on model serving, optimization, and monitoring to ensure speech capture is fast, reliable, and works both online and offline
• Partner with backend engineers to integrate speech models into the broader agentic pipeline that structures trials, tracks mastery, and drafts notes

What we're looking for

• Professional experience building and deploying speech recognition or audio ML systems in production
• Strong proficiency in Python and common ML frameworks (PyTorch or TensorFlow)
• Experience with ASR model architectures, fine-tuning, and evaluation (e.g., Whisper, wav2vec, or similar)
• Solid understanding of the practical tradeoffs between model accuracy, latency, and resource constraints, including on-device or offline inference
• Experience working with real-world, noisy audio data and building evaluation sets that reflect production conditions
• Comfort working in a domain with strict data privacy and compliance requirements

Nice to have

• Experience with healthcare, clinical, or other regulated-industry data
• Familiarity with HIPAA-compliant data handling practices
• Experience building systems that run partially or fully offline