Voice AI Prototyping Engineer & Researcher

Sage Care; Palo Alto (CA); On-site; USD 150,000 - 210,000; Full time

Job summary

Sage Care in Palo Alto is seeking a Software Engineer to own the AI pipeline from idea to evidence, staying close to research trends and the voice AI ecosystem.

You will design experiments, build rapid prototypes on production data, and turn investigations into team-usable artifacts with benchmarks, deep dives, or evidence-backed recommendations.

Collaborating with platform engineers, you will hand off validated ideas for production and help the team explain the stack choices six months in.

Qualifications

• 3+ years of software engineering or applied ML experience.
• Strong coding skills and ability to run experiments end-to-end without infra support.
• Hands-on experience with LLMs: prompting and evaluation.
• Experimental rigor: tests with controls and baselines.
• Experience teaching or sharing knowledge (talks, writings, open source).
• Intellectual honesty: reporting when ideas don’t work.

Responsibilities

• Track emerging voice AI techniques and identify what matters for us.
• Build rapid prototypes and test against real conversation data or in development.
• Run head-to-head evaluations of models, providers, and techniques.
• Turn investigations into team-usable artifacts: benchmarks or deep-dives.
• Collaborate with platform engineers to hand off validated ideas for production.

Skills

Years of software engineering

LLMs experience

Experiment design

Coding proficiency

Teaching/knowledge transfer

Integrity/ honesty in results

Education

MS/PhD in CS or ML

Job description

Sage Care in Palo Alto is seeking a Software Engineer to own the AI pipeline from idea to evidence, staying close to research trends and the voice AI ecosystem.

You will design experiments, build rapid prototypes on production data, and turn investigations into team-usable artifacts with benchmarks, deep dives, or evidence-backed recommendations.

Collaborating with platform engineers, you will hand off validated ideas for production and help the team explain the stack choices six months in.