gradstay
developed and iterated on a rental property platform, pivoting from estate agency to findhousing/yuzuu, a skyscanner for rentals with ai-powered chatbots and property matching (january 2024 - august 2025, full-time)
Work
gradstay landing page
built a landing page collaborating with the designer using reactjs, vite, and tailwind css which collected 100+ waiting list emails
reactjsvitetailwind csstypescript
tenant house profiles
built the first mvp feature testing the hypothesis of allowing tenants to create profiles for their houses in exchange for part of the commission when the house is rented out
reactjstypescriptfirebase authenticationfirestorefirebase cloud functionsgoogle analytics
tenant rental form
pivoted into a more traditional estate agency pipeline and built a form for tenants looking for rentals
reactjstypescriptfirestore
findhousing scraping pipeline
built scraping pipeline using firebase cloud functions and headless browsers with puppeteer to collect rental property data from the largest platforms in the uk
firebase cloud functionspuppeteerweb scrapingtypescript
property matching algorithm
constructed a sophisticated property matching algorithm to determine the probability that two listings from different platforms referred to the same property
typescriptpython
intelligent property search
intelligent searching of the scraped property database to fetch matches based on simple query
typescriptfirestore
rental value neural network
used python to create a neural network to predict rental value of a property built off the dataset of scraped properties
pythonmachine learning
findhousing landing page
created a new landing page for findhousing in collaboration with the designer that used reactjs, vite, and typescript which got user rental search details and whatsapp contact information
reactjsvitetypescript
whatsapp property chatbot
used the whatsapp cloud api and openai api to create a chatbot that would translate natural language into database queries and send the user a list of homes that matched their search
whatsapp cloud apiopenai apitypescript
yuzuu advanced chatbot
created a new version of the chatbot with more advanced onboarding capabilities and more specific natural language property searches, filtering homes based on text descriptions matched with user preferences, using the neural network rental predictor to advise users if their budget was too low for the area
whatsapp cloud apiopenai apitypescriptmachine learning
yuzuu web app
based on customer demand, created a new web app for yuzuu using reactjs, nextjs, typescript, and firebase that allowed for more traditional property searching, similar to the web version of skyscanner but for property rentals
reactjsnextjstypescriptfirebasefirestore
firestore query optimization
worked around the limitations of firebase firestore to optimize property fetching, including complex queries, sorting of query results, and support for lazy loading with query cursor
firestoretypescript
scrapingbee scraper upgrade
updated the scrapers to use scrapingbee for certain platforms
scrapingbeeweb scrapingtypescript
gradstay whatsapp onboarding bot
built a gradstay chatbot with whatsapp cloud api and openai api that onboarded users asking for their property rental search details, automating the estate agent task of collecting user details from the prospective tenant
whatsapp cloud apiopenai apitypescript
gradstay otp auth landing page
built a new landing page for gradstay with the designer that included an authentication flow using a custom sign-in flow of sending an otp to the user's whatsapp in order to sign up, signing up 750+ users
reactjstypescriptwhatsapp cloud apifirebase authentication
gradstay admin panel
built an admin panel to manage customer relationships with all the users, including seeing the user logs with the chatbot and their search details
reactjstypescriptfirestore
smart user-property matching
smart matching of users whenever a new property was onboarded, used to message all matching users
typescriptfirestore
landlord listing flow
built a new property listing flow for landlords, including smart address matching from postcodes, photo compression and uploads to firebase storage, and ai property description generation using photos and property details
typescriptreactjsfirebase storageopenai api
discord notifications
integration with discord to notify the company server whenever a new user or property is onboarded
discord apitypescript
whatsapp hypothesis testing
tested the key hypothesis of the findhousing lead magnet with 25+ test users via whatsapp, sending them houses that matched their criteria as soon as the scrapers found them
whatsapp cloud api
property partnerships
through key partnerships, onboarded 115+ properties on the database
stakeholder communication