ASR Engineer – Cloud and On‑Device Speech Transcription

About the role

You will own the transcription pipeline end‑to‑end at an early‑stage ambient‑intelligence consumer hardware startup, balancing latency, accuracy and reliability while working directly with the GM and Head of Product.

Key responsibilities

• Build and iterate on the cloud‑based ASR pipeline, from audio capture through post‑processing, at production scale.
• Own end‑to‑end ASR quality and reliability, delivering measurable improvements in latency, word‑accuracy and voice‑print reliability.
• Turn product feedback into shipped pipeline changes by collaborating with data, training, evaluation and deployment teams.
• Work closely with China‑based R&D, hardware and supply‑chain teams across time zones.

Required profile

• 3+ years experience building and tuning transcription or ASR pipelines in production, primarily cloud‑based.
• Proven ownership of the full ASR lifecycle: data preparation, model training, fine‑tuning, evaluation and deployment.
• Experience with latency‑sensitive or streaming audio pipelines and making trade‑offs based on real user feedback.
• Early‑stage or founding engineering experience, able to ship without large team support.

Required skills

• Core ML
• TensorFlow Lite (or similar on‑device ML frameworks)

What we offer

• Salary range $150,000 – $200,000 USD per year.
• Hybrid work model – three days per week in our San Francisco office.
• Visa sponsorship is not available for this role.