ML Engineer
About the role
ABOUT GUAC
Guac uses machine learning to do demand forecasting for grocers. We predict exactly how much of each product grocers will sell, on each day, at each of their stores.
Our mission is to put an end to the millions of tons of food that goes to waste every single day due to bad inventory planning.
We’re currently working with large supermarket chains around the world; we recently closed our seed round led by 1984 Ventures; and we are a team of 4 based out of New York.
We’re excited to make a huge impact in what is still a highly untapped space, and we’re looking for a machine learning engineer to join us on the journey!
ABOUT THE ROLE
At Guac, we’re working on a really intellectually engaging + difficult problem: predicting how people will behave in the future. As one of our first employees, you’ll have the unique opportunity to solve this hard ML problem and build groundbreaking algorithms and data pipelines.
The sorts of problems you’d be solving with us are:
ML Ops: we work with huge time series datasets, so we need to build out distributed data and training infrastructure. We also need to take in customer data and send out forecasts to our customers, so interfacing with a range of (sometimes antiquated!) APIs is part of the package.
AI research + modeling: we’re always looking for new ways to optimize our forecasts — whether that’s by finding ways to improve the explainability of our forecasts, discovering novel ways to ensemble different models or reconcile different levels of forecasts together, or implementing the latest model architectures.
Building data pipelines: one of our ‘secret sauces’ is the hundreds of customer-specific data points that we bring into our models - things like school term dates, sports fixtures, betting odds, Spotify listening data, and more. These help us to contextualize sales data and understand future consumer behavior better. We want to build out significantly more automated scraping tools to speed up and scale up our forecasts.
Structuring the world’s data: we work with a lot of messy, unstructured data — typos in product names, uncategorized SKUs, sparse datasets. Coming up with creative ways to encode this data is a key part of our forecasting (e.g. embeddings + dimensionality reduction + clustering for categorization).
ABOUT YOU
Proactive: we’re an early stage company, so a lot of the work you do will be building features/functionality from the ground up. Many architecture decisions and choices about which tech to use will be in your hands.
Curious: most of our biggest forecasting improvements and some of our most useful features have come from experimenting with unfamiliar data/technologies/approaches. You should love learning, researching, and testing things out.
Driven: we like to bite off a bit more than we can chew, and want to be surrounded by a team willing to work super hard to build something transformative for our customers.
WHY WORK WITH US
Intellectually challenging problems: the everyday problems you’ll be working on require a lot of thinking and creativity. If you derive a lot of fulfillment from solving hard problems, you’ll get that with us. You also won’t be pigeon-holed into one problem, product, or technology. The role will involve a range of ops-based, backend-based, ML-based, and data-based work.
Have a real-world impact: grocery food waste is a trillion dollar problem — and millions of tons of food are thrown out each year. Every percentage point improvement we make to our forecasting models translates to a direct reduction in those figures.
Be surrounded by an awesome team: we’re 100% in person and deeply value having a team that cares about solving interesting problems and has fun working.
Meet interesting customers: while most of the job will be dev-based, we like to spend time with our customers to understand their problems face-to-face. You’ll be able to come along for the ride and see the on-the-ground operations of the grocery industry.
OUR TECH STACK
Ops: GCP (GCS, Cloud Run, Compute Engine mainly), FastAPI, Dask
ML: PyTorch, XGB, OpenAI embeddings, SHAP, + more
Codebase: Python
About Guac
Guac uses machine learning to do demand forecasting for grocers. We predict exactly how much of each product grocers will sell, on each day, at each of their stores.
Our mission is to put an end to the millions of tons of food that goes to waste every single day due to bad inventory planning.