Grocery Run | Devpost

Store Choice Page
Map Page
Shopping Items
A snippet of online ordering!
Recipe Page

Inspiration

Let’s face it: getting your groceries is hard. As students, we’re constantly looking for food that is healthy, convenient, and cheap. With so many choices for where to shop and what to get, it’s hard to find the best way to get your groceries. Our product makes it easy. It helps you plan your grocery trip, helping you save time and money.

What it does

Our product takes your list of grocery items and searches an automatically generated database of deals and prices at numerous stores. We collect this data by collecting prices from both grocery store websites directly as well as couponing websites.

We show you the best way to purchase items from stores nearby your postal code, choosing the best deals per item, and algorithmically determining a fast way to make your grocery run to these stores. We help you shorten your grocery trip by allowing you to filter which stores you want to visit, and suggesting ways to balance trip time with savings. This helps you reach a balance that is fast and affordable.

For your convenience, we offer an alternative option where you could get your grocery orders delivered from several different stores by ordering online.

Finally, as a bonus, we offer AI generated suggestions for recipes you can cook, because you might not know exactly what you want right away. Also, as students, it is incredibly helpful to have a thorough recipe ready to go right away.

How we built it

On the frontend, we used JavaScript with React, Vite, and TailwindCSS. On the backend, we made a server using Python and FastAPI.

In order to collect grocery information quickly and accurately, we used Cloudscraper (Python) and Puppeteer (Node.js). We processed data using handcrafted text searching. To find the items that most relate to what the user desires, we experimented with Cohere's semantic search, but found that an implementation of the Levenshtein distance string algorithm works best for this case, largely since the user only provides one to two-word grocery item entries.

To determine the best travel paths, we combined the Google Maps API with our own path-finding code. We determine the path using a greedy algorithm. This algorithm, though heuristic in nature, still gives us a reasonably accurate result without exhausting resources and time on simulating many different possibilities.

To process user payments, we used the Paybilt API to accept Interac E-transfers. Sometimes, it is more convenient for us to just have the items delivered than to go out and buy it ourselves.

To provide automatically generated recipes, we used Cohere.

Challenges we ran into

Everything.

Firstly, as Waterloo students, we are facing midterms next week. Throughout this weekend, it has been essential to balance working on our project with our mental health, rest, and last-minute study.

Collaborating in a team of four was a challenge. We had to decide on a project idea, scope, and expectations, and get working on it immediately. Maximizing our productivity was difficult when some tasks depended on others. We also faced a number of challenges with merging our Git commits; we tended to overwrite one anothers’ code, and bugs resulted. We all had to learn new technologies, techniques, and ideas to make it all happen.

Of course, we also faced a fair number of technical roadblocks working with code and APIs. However, with reading documentation, speaking with sponsors/mentors, and admittedly a few workarounds, we solved them.

Accomplishments that we’re proud of

We felt that we put forth a very effective prototype given the time and resource constraints. This is an app that we ourselves can start using right away for our own purposes.

What we learned

Perhaps most of our learning came from the process of going from an idea to a fully working prototype. We learned to work efficiently even when we didn’t know what we were doing, or when we were tired at 2 am. We had to develop a team dynamic in less than two days, understanding how best to communicate and work together quickly, resolving our literal and metaphorical merge conflicts. We persisted towards our goal, and we were successful.

Additionally, we were able to learn about technologies in software development. We incorporated location and map data, web scraping, payments, and large language models into our product.

What’s next for our project

We’re very proud that, although still rough, our product is functional. We don’t have any specific plans, but we’re considering further work on it. Obviously, we will use it to save time in our own daily lives.