hero

Investing in visionary founders, transformational technology and emergent ecosystems for a new world.

Innovation Endeavors
Innovation Endeavors
69
companies
1,201
Jobs

Staff Applied Scientist - Forecasting

Afresh Technologies

Afresh Technologies

Remote · San Francisco, CA, USA
Posted on Friday, December 8, 2023

Afresh is on a mission to eliminate food waste and make fresh food accessible to all. Our first A.I.-powered solution optimizes ordering, forecasting, and store operations for fresh food departments in brick-and-mortar grocers. With our Fresh Operating System, regional and national grocery retailers have placed $1.6 billion in produce orders across the US and we've helped our partners prevent 34 million pounds of food from going to waste. Working at Afresh represents a one-of-a-kind opportunity to have massive social impact at scale by leveraging uncommonly impactful software – we hope you'll join us!

About the Role

The Prediction, Optimization, and Planning (POP) team builds Afresh's core replenishment technology. Our AI models are directly responsible for ordering millions of dollars of fresh inventory across the world every day. Fresh food ordering is an extremely complex high-dimensional decision-making problem. We face the complex challenges presented by decaying product, uncertain shelf lives, varying consumer demand, stochastic arrival times, extreme weather events, and tight performance constraints (to name a few). We tackle these problems with a mix of machine learning, large-scale simulation, and optimization technologies. Our team has presented work at multiple industry leading conferences, and we encourage team members to write and present their work publicly.

As an Applied Scientist at Afresh focused on forecasting, you will take your existing knowledge of machine learning and apply it to the challenge of demand forecasting. You will thrive in this role if you love the craft of neural networks in research and production environments: hyperparameter optimization, model tuning, feature selection, thorough error analysis, architecture selection, and deep understanding of the data you are modeling. We are looking for someone who is equally adept at gaining insights from 8 hours of data analysis or 8 hours of reading papers.

Your will research, implement, and rigorously validate improvements to our core forecasting system. This will include modeling seasonal demand, rare events, promotion cannibalization, and aggregating correlated demand across multi-echelon systems. Your work will be visible from day one, will make a substantial impact on decreasing food waste, and will lead to fresher, healthier produce for millions of people across the world.

  • You will work on improving the accuracy and calibration of the demand forecasting models that power our warehouse-level and store-level ordering products. You will also lead research development for new product and business challenges.
  • In your first 3 months, you will acquire a comprehensive understanding of our current forecasting system and modern neural network approaches to time-series forecasting. You will gain proficiency in our data manipulation, transformation, and forecasting tools, and you'll test an experimental improvement to our demand forecasting system.
  • By the end of your first 6 months, you will have proposed, implemented, and rigorously validated an improvement to our forecasting system.
  • By the end of your first year, you will have led the implementation of fundamental changes to our core forecasting system and led research into forecasting extensions (multi-echelon forecasting, rare event modeling).
  • We need to make optimal ordering decisions for millions of items for weeks at a time, and our system must be fault-tolerant to an extreme. Our partners rely on our system to order millions of dollars of inventory every week, and so your code must be rigorously validated, tested, and bug-proof.

Skills and Experience

  • 5+ years of industrial or academic experience building mission-critical machine learning systems.
  • Excellent communication and presentation skills. You should be able to explain complex mathematical ideas to product teams in plain English.
  • 5+ years of experience in neural network libraries (Pytorch/Tensorflow/JAX). Proficiency in Pytorch preferred, or willingness to transition from Tensorflow/JAX.
  • Fluent in programmatic data analysis using SQL and Pandas.
  • Excellent data visualization skills.

About Afresh

Founded in 2017, Afresh is working on the #1 solution to curb climate change: reducing food waste. By combining human insight and transformative technology, we're helping grocers provide fresher food to customers at more affordable prices.

Afresh sits at an incredible intersection of positive social impact, rocket ship financial growth, and cutting-edge technology. Our best-in-class AI research has been published in top journals including ICML, and we've raised over $148 million in funding from investors including former co-CEO of Whole Foods Market Walter Robb and Eric Schmidt's Innovation Endeavors.

Fresh is the past, present, and future of our food system – the waste we create today will impact our planet for years to come. Join us as we continue to build a vibrant, diverse, and inclusive team that embodies our company’s values of proactivity, kindness, candor, and humility.

Afresh provides equal employment opportunities (EEO) to all employees and applicants for employment without regard to race, color, religion, sex, national origin, age, disability, genetics, sexual orientation, gender identity/expression, marital status, pregnancy or related condition, or any other basis protected by law.

Here at Afresh, many of our employees work remotely provided that they reside in one of the following states: AR, CA, CO, FL, GA, IL, KY, MA, MI, MT, MO, NV, NJ, NY, NC, OR, PA, TX, WA, WI. However, there may be key roles that will require a candidate/employee to be local to our San Francisco, CA office. In which case this requirement will be included in the job posting details under "Skills and experience" for reference.