Straight to the employer's own application — no middleman.
The posting states 'Remote' with no geographic, timezone, or work authorization restrictions. The company is UK-based but the job description explicitly frames this as a growth role with no location requirement stated anywhere in the body. The 'Strong communication skills and fluency in English' requirement is satisfied by the target user. No visa sponsorship language appears, and the contractor/employee status is unclear but immaterial here since no visa is mentioned.
“Remote” Geographic scope — Job is explicitly remote with no geographic restriction stated in the posting.
“Strong communication skills and fluency in English.” Language — English fluency is the only language requirement; no additional languages are required.
“Competitive salary with the potential for equity options based on performance” Work authorization — No mention of work authorization requirements, visa sponsorship, or citizenship restrictions anywhere in the posting.
Dwelly — a UK-based, AI-enabled lettings and property management platform, that is growing through a roll-up strategy acquiring estate agencies. The company leverages two arms: i) acquiring existing letting agencies, effectively buying its highly sticky, recurring revenue-type landlords portfolios, and then ii) building a top-notch technology to automate tenant management, payments, and post-rental property maintenance. The company seamlessly integrates AI services to automate all business processes within brick-and-mortar real estate agencies, integrating them into a tech-enabled digital letting platform in two months to radically improve the user experiences and increase efficiency of the business.
We are launching Growth as a dedicated direction — everything that moves units on the platform and revenue per unit. You will be the first data person in it, owning the analytical agenda together with the commercial and product leads rather than servicing a queue of chart requests. A thought-partner role: bring hypotheses, argue about priorities, say when a plan won't work, then build the thing that settles it.
A note on the title. We have always asked our analysts for statistics, real programming, data pipelines and modest ML — we just never wrote it down. We are now naming the job the way the market names it. If "Data Scientist" means a research seat with a clean feature store handed to you, this isn't that. If it means going from raw messy data to a decision without waiting for anyone, it is exactly that.
A normal estate agency knows almost nothing about its own business: a CRM with names, a bank feed, and the memory of whoever has worked there longest. We are in a different position. Across a portfolio of agencies we hold years of conversations with landlords and tenants, every property management job with the full record of what went wrong, every payment and arrear, and thousands of hours of calls. Most of it is unstructured, which until recently meant unusable. With LLMs it is a feature store — and that changes the class of question we can answer. No agency on the island can do this, and few proptech companies can.
Why it's cool: a data asset nobody else in this market can replicate — years of conversations, jobs, payments and calls, and permission to point LLMs at all of it; experienced founders who have already walked this path before (built PIK Arenda), and now significant traction shown in the UK already; a large but compact and well-capitalized market of 20 thousand agencies on the island; Growth starts now, so you define the agenda and the metrics instead of inheriting legacy dashboards; and we are now on the way from 0 to 1, then there will be scaling 1 → N, so there's an opportunity to see how companies of different stages grow and develop.
Churn early-warning that names the cause, not just the risk. Combine arrears, job SLA breaches and tenancy events with intent and sentiment extracted from conversations and call recordings. Separate the landlord who is selling the flat from the one we lost through a botched boiler repair — different playbooks, one goes to retention, the other straight into the sales funnel — and put a pound figure on each cause so operations can prioritise honestly.
Turn share of wallet from a survey anecdote into a ranked call list. Landlords hold roughly 60 properties off-platform for every 100 they place with us. Estimate each landlord's hidden portfolio, rank by expected units won, then mine the resulting call recordings for why they said no — that is usually where the next product comes from.
Find the price sensitivity of the landlord base. Our acquired agencies charge wildly different fees. Reconstruct what is actually charged, estimate elasticity by segment, recommend the maximum defensible uplift — then hold yourself to your own churn forecast and correct the model. The same machinery prioritises the rent review backlog by expected pounds.
Underwrite Rent Guarantee off our own loss book. Probability of default and severity from our arrears and collections history, real pricing, eligibility rules, and monitoring that flags a deteriorating book early. We own the loss data, which is why we can build this and a broker can't.
Make growth experiments actually readable. Tenant-side products, opt-in payment flows, upsell paths and outreach sequences — designed with holdouts, power and an uplift estimate, not a before-and-after chart.
Own the Growth data layer. Pipelines from the platform database, payments, comms, call transcripts and PM jobs; the metrics tree everyone argues from; and an eval harness for the LLM extraction, because a classifier nobody has measured is a rumour.
Have real statistical depth: experiment design, causal inference without randomisation (diff-in-diff, matching, synthetic control), survival analysis, elasticity. You volunteer the caveat before someone else finds it.
Apply pragmatic ML — churn propensity, uplift, pricing, simple forecasting. For the current state we prefer a well-calibrated logistic regression in production to a gradient boosting model in a notebook.
Nice to have: have priced or underwritten a financial product — insurance, guarantee, credit — off your own loss data; enjoy being the first analyst in a direction and defining the metrics tree from scratch; and like talking to Managing Directors and agents and learning the domain first-hand, rather than only through the database.
What we offer is not a fancy office or a static workplace. Instead, this is solving one of worlds’ most complex problems in the largest consumer industry in the world (residential rentals), to improve the experience for >30% of households (>5M in the UK, and >100M including EU and US) that live in rental homes.
This is about disrupting the largest, most antiquated industry in the world, with one of the strongest operational and technical teams that exist in the UK and the EU. We work hard, and we shoot for extremely ambitious results. But we want people to be proud of what they’ve built and be able to look back and say one day “hell yeah, that was me that did it all”.
By applying for this position, you consent to the processing and storage of your personal data for recruitment purposes for up to 365 days, in accordance with our data retention policy and applicable data protection laws.
Straight to the employer's own application — no middleman.
Tell us and we'll mark Dwelly as a company that has actually hired in the Caribbean. It's the most useful thing on this board and it only exists because people report it.
I got hired here →
