Hard Problems, Real Impact
Whether you’re just beginning to tap into your data’s potential or you’re fine-tuning advanced data operations, we’re here to help. Our services span from fractional data science advisory to full-stack custom solutions.
Read on to explore how we’ve helped our partners solve their hard problems using data science and engineering.
How do you use telematics data to price insurance at a billion-dollar scale?
Pie Insurance needed to assess and de-risk the potential of a new telematics dataset before building a dedicated internal team. Echelon DS built a loss ratio model that formed the foundation of a business partnership with Ford to open up the $40B commercial auto insurance market. Echelon DS remains engaged with Pie, contributing to their core underwriting.
How do you identify cancer nodules from X-rays with superhuman performance?
Imidex had invested heavily in building a robust dataset of X-rays; however, their team struggled to develop a model that would meet FDA standards. Echelon DS built a computer vision lung cancer detection model that surpassed FDA medical device standards, ultimately leading to 510k FDA approval for Imidex’s VisiRad™ XR product.
How do you pull valuable, relevant information from hundreds of public web pages every week?
IDI was spending hundreds of hours manually parsing data on potential deals. Echelon DS developed a custom app that scrapes 100+ unique government entity websites and sends weekly AI-generated summaries of all relevant newly published content, reducing IDI’s time-to-value on public information and freeing their team for higher-value initiatives.
How do you forecast restaurant demand in real time to staff locations efficiently?
Lineup wanted to reduce labor costs across its restaurant locations by optimizing scheduling to more accurately reflect demand. By combining endogenous sales data with exogenous features like weather, events, holidays, and TV schedules, Echelon DS improved demand prediction accuracy by 40%, resulting in material labor cost savings across Lineup’s portfolio.
How do you instantly price waste collection services for every US ZIP Code? (>44M Combinations)
In Rubicon’s early days, its growth was bottlenecked by long delays from having to manually provide price quotes to customers. Echelon DS built a nationwide pricing engine that reduced the average quote time from over a day to less than a minute. This unlocked exponential growth for Rubicon’s core business, paving the way to its IPO five years later.
How do you make real-time strategy recommendations when every second counts?
Created decision-making software for racing teams that won multiple NASCAR championships. Software directly drove at least three race wins when crew chiefs followed strategy recommendations made in the final stages of the race. Acquired in 2022 by General Motors (NYSE: GM).
What’s the mathematically optimal way to bet on baseball?
Faced with rising student loans and a strong 2005 Red Sox roster, Echelon DS’s cofounders created BBHedge, a series of predictive baseball models that returned 33,000% in one year. BBHedge was forcibly shut down after its rookie season due to an online gambling ban signed into law by Congress.
How do you create a singular source of truth for data across your enterprise?
A national janitorial and landscaping company was struggling with fragmented data systems across its enterprise. Echelon DS developed a robust data warehouse with a universal schema to streamline the company’s data management capabilities. This unlocked immediate real-time, unified reporting while laying the foundation for AI-driven automation and analysis.
How do you use truck cameras to locate garbage cans in every neighborhood in America?
Rubicon (NYSE: RBT) needed computer vision models to precisely locate garbage cans on collection routes, including those not yet explored. Echelon DS built production-ready computer vision models that formed the bedrock of this strategy, which evolved into Rubicon’s Fleet Technology Business Unit and sold for ~$100M in 2024.
How do you automatically identify every unique vehicle that enters and exits a parking lot?
FlashParking partnered with Echelon DS to build and launch the parking and smart city application industry's first ever general-use unique vehicle recognition system. FlashParking plans to scale this product, built on the latest machine learning models and trained on generative Al data, to over 16,000 locations across North America.
A problem-solving partner that prioritizes results.
Data science is core to modern business strategy, not an afterthought. We work to understand our clients’ interests deeply and prioritize projects that deliver results.
Interested in working with us? Contact us to learn more about how we can support your business today.