工程主管

Most infrastructure problems worth solving are either scientifically hard or engineered at scale, rarely both at once. We are the exception. Over the past 18 months, we have built a physics-based model of the global freight network from scratch: 1,400 ports mapped at the berth level, 28 chokepoints, 140 carriers, 7,500 vessels, and 54 million customs manifests that give us per-account, per-lane intelligence no competitor can replicate. The science is done. We are currently building the product layer: the APIs, the agent pipelines, the model serving infrastructure, and the enterprise interfaces that make this intelligence usable for the supply chain teams who need it. That is the engineering problem we are hiring for, and it is a genuinely hard one: building a production-grade agentic system that ingests real-time signals across a global freight network, runs inference at the right latency for operational decisions, and surfaces the right action to the right person at the right moment.

8-14 years of experience, with meaningful time at a startup or high-growth tech company - not a long enterprise stack career
Has shipped backend systems to production and led or been a core member of a small, high-output engineering team (3-10 people)
Has seen how agentic AI infrastructure works - LLM orchestration, model serving, agent pipelines - even if they haven't architected one from scratch
Comfortable across the full backend stack: APIs, cloud infrastructure (AWS/GCP/Azure), databases, and system design
Strong opinions on code quality, shipping cadence, and building for reliability from day one
Ideally a founding engineer from a prior startup, looking for their next mission - not someone seeking a stable senior role

We are the exception. Over the past 18 months, we have built a physics-based model of the global freight network from scratch: 1,400 ports mapped at the berth level, 28 chokepoints, 140 carriers, 7,500 vessels, and 54 million customs manifests that give us per-account, per-lane intelligence no competitor can replicate. The science is done. We are currently building the product layer: the APIs, the agent pipelines, the model serving infrastructure, and the enterprise interfaces that make this intelligence usable for the supply chain teams who need it.