AI Engineer, Intake & Summaries

About Zaras

Zaras is building the operating layer for the trades — one record that carries a service business from first call to deposit cleared. Our team is small, hands-on, and field-first. We operate with a flat structure. All employees are expected to contribute directly to the work. Leadership is given to those who show initiative and consistently deliver.

About the role

Build the earn-its-keep AI surface — inbound call intake, follow-up drafting, and summarization that a dispatcher actually trusts. You will define the eval set and own the false-positive rate.

Responsibilities

• Ship the call-intake pipeline from transcript to job record.
• Build the eval harness and keep regressions visible.
• Tune prompts, retrieval, and fallbacks for noisy, real-world audio.
• Sit with dispatchers and convert their corrections into training data.

Basic qualifications

• 3+ years shipping production ML or LLM features.
• Strong Python or TypeScript; comfort with both is a plus.
• Built an eval set and used it to drive a measurable improvement.
• Pragmatic about latency, cost, and fallbacks.

Preferred skills and experience

• Voice or audio pipelines in production.
• Retrieval-augmented systems at scale.
• Experience with field-collected or dirty real-world data.
• A bias toward removing features that don’t earn their keep.

Zaras is an equal opportunity employer.