Senior AI Builder

Morning is looking for a Senior AI Builder to take AI from a "promising demo" to real production systems that customers rely on — and that continue to work well over time.
Our R&D group is a team of 12 engineers and one architect, spanning backend to mobile, building the accounting platform businesses actually run on every day.

What you'll do

Build and ship LLM-powered AI capabilities to production- agentic flows, retrieval over our data, extraction from real-world documents, and more. Solutions need to be accurate, measurable, cost-efficient, and rollback-ready.
Build and lead the evaluation process — continuously measure the quality of models and solutions, and be able to answer clearly: is the new version actually better than last week's?
Own whatever it takes to get the solution into production- from AWS Lambda to Vue and PHP. You don't need to be an expert in every technology, but you should be comfortable getting into the code and solving problems wherever the work actually lives.

Why this is a real challenge — and why it's worth it
Our customers trust us with their financial data. There's no room for "almost right" when it comes to the answers our systems provide, and we operate in an environment with significant regulatory requirements- including PCI-DSS, GDPR, and Israeli privacy law.
The goal is to take AI capabilities, turn them into real products people can trust, and see them reach customers- at a company small enough that your work can ship this quarter, not next year.

Requirements:

5+ years of experience building and operating production systems — real hands-on experience building software, shipping it to production, and operating it over time.
Significant hands-on experience building with LLMs- including dealing with challenges like retrieval quality degrading, agent loops running out of control, token costs spiking, and reliability and accuracy issues.
Strong TypeScript or Python skills and comfort with AWS serverless.
The engineering judgment to tell a stakeholder: "This isn't an AI problem."

Advantages:

Experience with Databricks/Spark, Vue 3, fintech or another regulated domain, and significant experience with AI-powered development tools and agentic development.