AI / Machine Learning Engineer
13 hours ago
British Columbia, Canada
Listifyy
Remote
Full-time
Free with email or Google
Save this job and keep your search organized
Create a free account to save jobs, create alerts and return to this listing from your dashboard.
Free with email or Google
By continuing, you agree to our Terms & Privacy Policy.
Listifyy is a Vancouver-based Canadian real-estate platform. We want buyers, sellers and agents to see a realistic estimate of what a home is likely to sell for, not just its list price.
About the role
This is a part-time, project-based, remote role for candidates based in Canada. You will lead the modelling work for our home-valuation engine, a project we expect to build in stages over roughly one to three years. The models estimate the likely sale price of a Canadian home, together with a price range we can stand behind. The hard part is the data. Sold-price records are plentiful in large cities and thin in rural areas, and listings often miss details like floor area. The models must stay accurate, and honest about their uncertainty, where data is scarce. Engagement
• Part-time and remote (Canada)
• Hourly or fixed-price per project stage, to be agreed
• Stage-based: each stage starts from what the previous one proved
• Long-term potential for the right person
What you'll do
• Turn raw listing and sold-price data into clean, tested training sets
• Build price models on tabular data, including models for missing features
• Produce price ranges that hold up region by region
• Measure error by province, urban/rural area and price band, and test on later sales
• Plan each piece of work as an experiment, run it reproducibly and write up what it showed
• Later, help deliver model outputs to the Listifyy website and keep them monitored Must have (required)
• Based in Canada and legally authorized to work in Canada
• Strong Python (pandas, scikit-learn)
• Hands-on regression on tabular data with gradient boosting (XGBoost, LightGBM, CatBoost)
• Sound validation habits: cross-validation, time-based splits, and avoiding data leakage
• Experience handling messy, missing or sparse data
• Clean, version-controlled code with tests (git, pytest)
• Clear written English: you can explain a result, and its limits, to non-specialists
• Careful documentation of experiments: hypothesis, method, results and failures Nice to have
• Real estate, pricing or geographic data
• Prediction intervals (quantile regression, conformal prediction)
• Experiment tracking (for example MLflow or DVC)
• Putting models into production and monitoring them
• Knowledge of the Canadian housing market
• Small-team or early-stage experience How we work We treat the work as research. Hypothesis first: each experiment states its question and plan before it runs. We write up failed experiments as carefully as successful ones. Every reported result is reproducible from its code version and data snapshot. You work in a Listifyy GitHub account, and listing data containing personal details stays out of code, logs and documents. We work remotely, with regular written updates and check-ins. To apply Please include: 1. A model you built on tabular data, how you validated it and what went wrong 2. How you would approach a price model where data is thin in some regions 3. Your availability (hours per week) and expected hourly rate or project pricing 4. Any questions about the project
About the role
This is a part-time, project-based, remote role for candidates based in Canada. You will lead the modelling work for our home-valuation engine, a project we expect to build in stages over roughly one to three years. The models estimate the likely sale price of a Canadian home, together with a price range we can stand behind. The hard part is the data. Sold-price records are plentiful in large cities and thin in rural areas, and listings often miss details like floor area. The models must stay accurate, and honest about their uncertainty, where data is scarce. Engagement
• Part-time and remote (Canada)
• Hourly or fixed-price per project stage, to be agreed
• Stage-based: each stage starts from what the previous one proved
• Long-term potential for the right person
What you'll do
• Turn raw listing and sold-price data into clean, tested training sets
• Build price models on tabular data, including models for missing features
• Produce price ranges that hold up region by region
• Measure error by province, urban/rural area and price band, and test on later sales
• Plan each piece of work as an experiment, run it reproducibly and write up what it showed
• Later, help deliver model outputs to the Listifyy website and keep them monitored Must have (required)
• Based in Canada and legally authorized to work in Canada
• Strong Python (pandas, scikit-learn)
• Hands-on regression on tabular data with gradient boosting (XGBoost, LightGBM, CatBoost)
• Sound validation habits: cross-validation, time-based splits, and avoiding data leakage
• Experience handling messy, missing or sparse data
• Clean, version-controlled code with tests (git, pytest)
• Clear written English: you can explain a result, and its limits, to non-specialists
• Careful documentation of experiments: hypothesis, method, results and failures Nice to have
• Real estate, pricing or geographic data
• Prediction intervals (quantile regression, conformal prediction)
• Experiment tracking (for example MLflow or DVC)
• Putting models into production and monitoring them
• Knowledge of the Canadian housing market
• Small-team or early-stage experience How we work We treat the work as research. Hypothesis first: each experiment states its question and plan before it runs. We write up failed experiments as carefully as successful ones. Every reported result is reproducible from its code version and data snapshot. You work in a Listifyy GitHub account, and listing data containing personal details stays out of code, logs and documents. We work remotely, with regular written updates and check-ins. To apply Please include: 1. A model you built on tabular data, how you validated it and what went wrong 2. How you would approach a price model where data is thin in some regions 3. Your availability (hours per week) and expected hourly rate or project pricing 4. Any questions about the project