Pearl is shaping the future of dentistry with a suite of AI solutions developed to establish higher standards of quality and care for patients worldwide. Since 2019, our team has engineered FDA-cleared computer vision capabilities for the interpretation of 2D and 3D dental imagery — industry-leading capabilities which clinicians, practice owners, labs, and insurers use to elevate the efficiency, accuracy, and consistency of dental care around the world.

At Pearl, we believe AI isn’t just what we build—it’s how we build. We’re looking for engineers who embrace AI-powered development tools to amplify their output and ship better software, faster.

The Role

Dental insurance is messy. Every payer has different rules, different data formats, and different ideas about what they’ll cover and when. We’re building AI that cuts through that complexity so dentists can focus on patients instead of navigating insurance. Our product sits at the intersection of clinical decisions and insurance data — and the quality of our recommendations lives or dies on the quality of that data.

The data is high-cardinality, spans multiple domains, and comes from vendors with varying levels of reliability. Fields are missing. Values contradict each other. Connections go stale. We want to catch these problems before our customers ever feel them — and unlock new value by trusting the data enough to build on it.

We need someone who owns this problem end to end. Not as a side project for the engineering team, but as the job.

Think of it as detective work backed by engineering. You’ll trace bad recommendations back to their source, build systems that catch problems before customers do, and hold vendors accountable for the data they send us. Some days that’s writing code, some days it’s writing a report for a vendor call — but the mission is always the same: make the data right.

This role has a direct line from your work to the product getting smarter. Every data problem you solve makes our recommendations more trustworthy — and dentists notice.

What You’ll Do

Analyze incoming vendor data across every domain we integrate with to identify omissions, inaccuracies, and inconsistencies — the data is complex, and the problems are often subtle
Build automated data quality monitoring and alerting — design and implement systems that catch problems before they reach our customers
Measure and improve the precision and recall of our recommendations — quantify the impact of data issues, prioritize by product impact, and track improvement over time
Own vendor data quality relationships — work directly with data providers through structured reporting, regular syncs, and formal SLAs to drive upstream fixes
Proactively detect stale or broken data connections and find ways to reduce the manual overhead required to keep them healthy
Write application code to validate, transform, and reconcile incoming data — not just queries, but programmatic detection and correction systems
Communicate findings to engineering, product, and leadership — translate data problems into product impact that drives prioritization

What You'll Need to Succeed

Senior-level engineering skills — you’re a strong programmer, not an analyst who can script. You can build production systems, not just notebooks
Node.js proficiency — TypeScript experience is a plus. Our stack is Node, AWS Lambda, PostgreSQL, with Snowflake available for analytical workloads
Strong SQL skills — you’ll live in PG and Snowflake, exploring large datasets to find patterns in what’s missing or wrong
Product mindset — you think about data quality in terms of “what does this do to the user’s experience,” not just “is this field populated.” You’ll need to understand our product deeply to do this job well
Self-directed problem finder — you don’t wait to be told what’s broken. You dig, you notice, you quantify, and you propose the fix
Comfortable owning vendor relationships — you’ll be on calls with data providers, writing up findings, pushing for fixes. This requires clarity, persistence, and professionalism
Enthusiastic about agentic software development — you actively use tools like Cursor and Claude Code to multiply your output. This isn’t optional or aspirational; it’s how you work today

Nice to Have

Healthcare or insurance domain knowledge is a plus, not a requirement — if you’ve worked with complex, regulated data where accuracy has real consequences, great. If not, you need the curiosity and capacity to learn fast