AI Tutoring Dialogue Quality Engineer

Maltex AI

Remote · $130k – $160k / year · Full-time · Posted 4 months ago

This role is no longer accepting applications

Don't miss the next role like this

Create a free profile with your skills. We match you to new AI roles as they open, and companies hiring can contact you directly.

Create free profile, 2 minutes

Similar open roles you can apply to now:

See all open AI jobs

This role is no longer accepting applications

Don't miss the next role like this

Create a free profile with your skills. We match you to new AI roles as they open, and companies hiring can contact you directly.

Create free profile, 2 minutes

Similar open roles you can apply to now:

See all open AI jobs

About the role

An AI tutor that explains a concept clearly once is useful. An AI tutor that notices when the student is confused, adapts its explanation approach, provides targeted practice, and knows when to simply give the answer versus guide the student to discover it — that is genuinely transformative for learning outcomes. The difference between those two things is dialogue quality. Our adaptive learning platform serves 800,000 K-12 and university students. The tutoring dialogue system generates explanatory responses, hints, and practice problems in real time. We have 3 years of student interaction data with learning outcome labels — we know which dialogue patterns actually produce learning. But the models that generate dialogue are not yet informed by that signal. You will build the dialogue quality evaluation and improvement pipeline: developing learning outcome predictors from interaction features, creating quality classifiers that identify dialogue turns likely to confuse rather than clarify, and designing the RLHF-style training signal that uses learning outcomes as the reward for dialogue quality optimization.

Fully remote.

Responsibilities

Build learning outcome predictors from student interaction and assessment data
Develop dialogue quality classifiers identifying confusion-inducing response patterns
Design RLHF-style training signal using learning outcomes as dialogue quality reward
Create evaluation framework measuring dialogue quality against student learning gains
Build A/B testing infrastructure for dialogue strategy experiments
Collaborate with learning science researchers to validate model-based quality signals

Requirements

5+ years in NLP engineering with dialogue systems or conversational AI experience
Experience with RLHF, preference learning, or reward model design
Strong Python and PyTorch skills with LLM fine-tuning experience
Ability to work with education domain experts on pedagogy-informed system design
Interest in education technology and learning outcomes