Job Opportunities API

Nobody calls a logistics coordinator to say things went well. They call because a $100,000 shipment of clinical trial medication has been sitting in customs for three days and nobody can explain why. Because a biologic therapy arrived outside its temperature window and the treatment has to start over. Because the patient is waiting and nobody in the supply chain can give a straight answer about the shipment’s location. This is the problem we’re solving. We’re a year in, tracking over 500,000 shipments (including for Fortune 100 customers), and we’ve already cut manual ops effort in half. General Catalyst, Eclipse Ventures, and Virtue led the seed. The angels are all former or current operators who've spent careers tracking down shipments themselves, such as Head of Logistics at Bristol Myers Squibb, CMO of Cardinal Health, President of Novo Nordisk US, President of UPS Air, and CEO of Uber Freight. They know the market is as large as the problem is broken. What we’ve built works, and the most interesting problems are still ahead of us. - Olivier, Co-founder & CEO THE ROLE Founding data scientist. The platform sees more about a shipment than anyone else in the supply chain does. Your job is to turn that into prediction: models that flag the customs hold, the temperature excursion, and the silent carrier before the issue reaches a patient. This is the Predict in our See/Predict/Act framework, and it is the part customers cannot get anywhere else. You'll own it end to end: the models, the data and feature infrastructure they run on, and the experiments that prove they work. You'll define how we measure a model, not just build one. Every person at ZoomLogi talks directly to customers, and you will too. The ops manager who tells you which exceptions actually hurt is the best feature-selection input you'll ever get. WHAT YOU'LL BUILD - The risk and prediction models. The platform ingests a continuous stream from carriers, forwarders, IoT sensors, weather, and flight data, stitched into a unified picture of every active shipment. You'll refine the models that scan that picture for risk and flag issues before they cascade, across hundreds of carriers, lanes, and sensor types. The hard part is the long tail and the asymmetry: a missed signal is measured in patient outcomes, and a noisy one trains operators to ignore you. Getting precision and recall right on real shipments is the work. - Dynamic re-routing optimization. Detecting a problem is half the job. The other half is deciding what to do about it: which alternate routing, carrier, or mode actually saves the shipment, and whether the fix is worth its cost and risk. You'll build the optimization that turns a flagged exception into the best recoverable plan, balancing transit time, temperature exposure, cost, and the hard constraints a healthcare payload carries. This is the Act in See/Predict/Act, and it is where a model stops being an alert and starts being a decision. - The data and feature infrastructure. A model is only as good as what feeds it. You'll own the pipelines and feature layer that turn 50+ messy external sources into model-ready signal at low latency. Some sources claim a shipment is delivered while the GPS shows it still in transit. Some go silent without warning. Building features that hold up against that, and that the next model can reuse, is yours. - The experimentation system. How do we know a model is actually working on live shipments? You'll build the offline evaluation and online experimentation that answers that: backtesting against real outcomes, measuring lift, and catching drift before customers do. You'll define the metrics the rest of the team trusts. WHO THRIVES HERE - You've shipped a model to production and watched what it did there. You have a specific recent example of something you built, owned, and put in front of real decisions, and you care about what happened after it shipped. Notebook-to-nowhere work i