Operations ML Engineer

Legitmark is building the trust layer for resale. We believe that authentication should be fast, accurate, and accessible to everyone, from individual sellers to the world's largest marketplaces. Our team of experts and engineers authenticate millions of items annually, helping buyers shop with confidence and sellers maximize value.

As an AI/ML Engineer on our Operations team, you'll build intelligent automation that makes our operations dramatically more efficient. This role sits at the intersection of machine learning, computer vision, and workflow orchestration. You'll identify bottlenecks, automate repetitive decisions, and build systems that route work to the right people at the right time. This is a greenfield opportunity. You'll design and build our operations AI infrastructure from the ground up, working closely with the team that processes items daily. The goal is to dramatically reduce manual work: automating image quality checks, pre-filling forms with high-confidence predictions, intelligently assigning tasks, and flagging exceptions before they become problems. This is a high-ownership role. We're a small team moving fast, which means you'll take projects from idea to production with minimal handoffs. You'll work directly with the founding team and have real input into what we build, not just how. We expect you to leverage modern AI-assisted development tools to ship quickly and maintain high quality. This role is remote-first with flexible hours. We work asynchronously across time zones and value output over presence.

We're an AI-native engineering team. We use Cursor with Claude as our primary development tools and expect engineers to be proficient with AI-assisted workflows. That includes debugging, code generation, refactoring, and rapid prototyping. If you're still writing every line manually, this probably isn't the right fit. This doesn't mean we want prompt-and-pray engineers. You need deep technical fundamentals to know when the AI is wrong, architect systems correctly, and debug complex production issues. AI tools make good engineers faster; they don't replace engineering judgment. We're pragmatic about ML. We reach for foundation models and APIs before training custom models, optimize for time-to-value, and only build complex systems when simpler approaches fail. You should have strong opinions about when to fine-tune vs. prompt engineer vs. build from scratch.