Software Engineer – Platform & Operations

About Us MarketLeap is building the first e-commerce operating system — an AI-driven platform that lets direct-to-consumer brands scale effortlessly across global marketplaces, the same way Amazon transformed retail. Since launching in 2022, we've helped brands sell in 30+ countries and raised $10.7M from investors including Smedvig Ventures, Expon Capital, Notion Capital, Motier Ventures, and Kima Ventures. Today's brands lose most of their time and margin just keeping up with the operational complexity of selling on marketplaces — sourcing, pricing, logistics, inventory, compliance — instead of building their products. We're solving that by turning e-commerce operations into software: automating workflows end-to-end, then layering specialized AI agents on top that perceive, decide, and act across the entire value chain — catalog, pricing, marketing, inventory, and logistics. We're a small, technical team tackling a genuinely hard systems problem: building composable, API-first infrastructure that connects to dozens of external marketplaces, suppliers, and carriers, generating the data exhaust needed to train and deploy autonomous agents at scale. If you want to work on AI that actually runs a business rather than just answers questions, and see your code make real, measurable impact on real brands' revenue — this is that kind of place. The role You will own the operational backbone of the platform: the workflows that move every order from click to doorstep and back. Ordering and fulfilment run across multiple channels today, and the next stage automates the hard parts: returns, replacements, cancellations, refunds, customer inquiries, and invoice generation, end to end. It also brings new fulfilment networks online: merchant-fulfilled (FBM) workflows and third-party logistics (3PL) integrations, supported end to end. Alongside the workflows, you'll own the shared platform every service and every AI agent stands on: a unified workflow engine, a global notification service, admin tooling, CI/CD, observability, tenant isolation, and backup/DR. This is the seat for an engineer who likes both sides of the coin: business-critical workflow correctness (an order is money, a refund is money twice) and the platform engineering that makes twenty services feel like one system. What you'll do Build and automate multi-channel ordering and fulfilment workflows: orders, returns, replacements, cancellations, refunds, inquiries, invoicingBring new fulfilment networks online end to end: merchant-fulfilled (FBM) workflows and 3PL/carrier integrationsDesign the unified workflow engine these flows run on, with retries, idempotency, and full auditability as first-class featuresShip shared platform services: global notifications, admin tools, and the primitives other engineers and AI agents build onRaise the platform's reliability floor: CI/CD, regression testing, observability, tenant isolation, infrastructure as code, backup/DROwn your services in production: Java (Quarkus) microservices on AWS serverless, plus the occasional React screen for the operators who use your workflows dailyExpose workflow actions and events to our agent platform with human-in-the-loop controls, so agents can safely operate what you build Our stack Services: Java (Quarkus) microservices on AWS serverless (Lambda, SQS, DynamoDB, S3), infrastructure as code with CDK (TypeScript), CI/CD on CircleCIFrontend: React/TypeScript internal platform used daily by our brand operatorsData & AI around you: a lakehouse data platform (Airbyte, Prefect, dbt, Athena on Iceberg) and an AI agent program consuming the events and actions your services exposeThe platform layer is early: the workflow engine, notification service, and admin tooling are yours to shape rather than inherit How we work We ship, then sharpen. V1 goes live in weeks, deliberately simple, and earns its sophistication with real usage. A workflow running in production beats an elegant state machine on a whiteboard.You own outcomes, not tickets. "Returns and refunds run without a human touching them" is the deliverable; how you get there is your call.Our customer is a brand's P&L. Every workflow you automate gives a brand back margin and time: a refund processed correctly, an invoice generated on time, an order that never needed a human. If a feature can't move a brand's number, we don't build it.Small team, no handoffs. Up to 10 engineers, the CTO in the room, and founders one Slack message away. Decisions in days, not quarters.Disagree, commit, measure. We argue with data, decide fast, and let results settle the debate. What we're looking for 5+ years building production backend services, with strong Java (or another JVM language) and comfort in TypeScriptYou've built business-critical workflow or transactional systems where correctness is money: ordering, payments, fulfilment, billing, or similarPlatform instincts: you notice the missing shared service, the flaky deploy, the unobservable failure, and you fix the class of problem rather than the instanceYou've owned services end to end in production, and event-driven serverless architecture is familiar ground or an easy reachBonus: workflow engines or orchestration systems, e-commerce or logistics operations, AWS CDK/Terraform, multi-tenant SaaS, exposing tools/actions to LLM agents Why this is a rare seat Every workflow you automate is felt by every brand on the platform the same week it ships, and the platform layer you build becomes the ground every future service and AI agent stands on. Operational excellence here isn't a support function; it's the product.