Senior AI Product Engineer

Just Right Reader is a forward-thinking curriculum provider committed to advancing literacy for students from Birth through 12th Grade. Our comprehensive resources are designed to support all learners, including upper-grade students requiring additional intervention to develop foundational skills. Grounded in the Science of Reading, our high-quality materials and tools equip educators with the support they need to accelerate achievement and foster lifelong readers.

Position Overview:

Just Right Reader works with school districts and state education agencies to get kids books at the right reading level. Our platform is TypeScript, Next.js and Postgres, and it connects to the CRM, ERP and internal systems that run the business.

We're adding AI to the product. We have a list of projects we want to build and we haven't started on any of them, so we're hiring an engineer to do that work.

Key Responsibilities:

The projects are already scoped, so you wouldn't be coming in to figure out what we should build. You'd be building it, and making the technical decisions along the way: how we call models, what happens when the output comes back wrong, what we do about cost and latency, and how we test any of it.

That last part matters more than it might sound. Right now we'd have no way to tell whether a change to a prompt or a model made things worse, and we'd want that in place early rather than after something goes out badly.

Because we work with K-12 students, privacy and data handling come up in every design conversation instead of at the end of one. Recommendations also have to be defensible, and one that's wrong but sounds confident causes us more trouble than not making one.

You'd also do regular product work. We're a small team, so there's Next.js, schema design and integration work alongside the AI projects.

Qualifications:

You've built an AI product from scratch and shipped it to real users. We're specifically after someone who made the architecture decisions rather than building features on top of a system another team maintained, since that's the situation here. If every AI project you've worked on already had a gateway, an eval setup and a cost dashboard in place when you arrived, this will be a harder jump than it looks.

We'd want to hear about at least one thing you built that people depended on, and we'll ask a lot of questions about it.

Engineering

• Senior-level experience as a full-stack or product engineer on systems that were genuinely in production, with real users and real consequences when something broke.
• Strong TypeScript. Nearly everything you'd write here is TypeScript, across both the application and the services around it.
• Next.js in production rather than familiarity with it. You should be comfortable with the App Router, server components, API routes, and the places where Next.js makes things harder than they need to be.
• Strong backend skills: API design, server-side business logic, and the kind of debugging where the bug turns out to be three systems away from where the error showed up.
• Strong PostgreSQL. Query design and optimization, schema and data modeling, and enough experience to know when the ORM is helping and when you should write the SQL yourself. A fair amount of the work here is SQL-heavy.
• Third-party API integration where the interesting part was the failure cases. Retries, idempotency, rate limits, partial failures, and vendors behaving differently in production than their docs suggested.
• AWS at the application level - S3, SQS, Secrets Manager, ECS/Fargate, Lambda, RDS, DynamoDB. We care more about whether you've debugged something running on them than whether you can list the services.

AI specifically

• You know the ways LLMs fail in production: malformed structured output, schema drift, latency spikes, cost surprises, refusals, and answers that are wrong but sound certain. More to the point, you know what to build so that users don't have to deal with any of it.
• You've had to evaluate quality somehow. Golden sets, scored runs, human review, regression checks before a deploy - whatever form it took, you've answered the question of whether a change made things better or worse with something other than a feeling.
• Judgment about when a model is the wrong tool. A fair amount of what looks like an AI problem turns out to be a rules or SQL problem with a better interface, and we'd rather ship the boring version when that's the right answer.
• Some sense of what running this costs and what to do about it.

Working here

• We're small, so you'd be talking directly with the people in operations, customer success and fulfillment who use what we build. Being able to have those conversations well matters about as much as the code does.
• Comfort treating internal tooling as a normal part of the job rather than a distraction from the interesting work.

Useful but not required:

• Retrieval systems in production, eval tooling you've built or used seriously, Prisma, Terraform or general infrastructure literacy, Python, previous K-12 or FERPA work, and experience with systems like NetSuite, HubSpot, Shopify or Slack.

We're not doing model training, fine-tuning or research, so this isn't the right role if that's what you're after.

Why Join Us:

At Just Right Reader, we're transforming how children learn to read. You'll be part of a passionate, growing team committed to educational equity, literacy innovation, and joyful learning.

• Mission-driven company focused on improving literacy and education outcomes.
• Collaborative and innovative work environment with opportunities for growth and development.
• Competitive salary and benefits package.