About the role
Description
We're looking for a Senior or Staff Machine Learning Engineer to build machine learning at UNiDAYS from the ground up. This is a hands-on role for someone who turns business questions into measurable ML problems: finding the right data, setting up the processes to clean and label it, defining how we measure success, and building the training and evaluation harness that lets us iterate safely.
We care more about getting the end-to-end loop right than about deep model research. You'll be pragmatic about when to use classical ML, when large language models are the better (or cheaper) tool, and when a simple heuristic is enough. You'll work closely with Product, Data and Platform, and over time help make ML something other engineers can build on
Requirements
We're looking for a Senior or Staff Machine Learning Engineer to build machine learning at UNiDAYS from the ground up. This is a hands-on role for someone who turns business questions into measurable ML problems: finding the right data, setting up the processes to clean and label it, defining how we measure success, and building the training and evaluation harness that lets us iterate safely.
We care more about getting the end-to-end loop right than about deep model research. You'll be pragmatic about when to use classical ML, when large language models are the better (or cheaper) tool, and when a simple heuristic is enough. You'll work closely with Product, Data and Platform, and over time help make ML something other engineers can build on
Requirements
- Taken ML from problem framing to production, and measured its impact with real experiments rather than offline scores alone.
- Built data pipelines and training datasets on messy product data, avoiding data leakage and using time-based train/test splits.
- Designed evaluation frameworks and A/B tests, including spotting when offline gains did not carry over online.
- Strong Python and software engineering practice: tests, CI, deploying and monitoring models, not only notebooks.
- Used LLMs in production with an eye on cost, latency and quality, and understand the pitfalls of using an LLM to judge outputs.
- Comfortable working with ambiguity and starting from a blank page, choosing the simplest thing that works and iterating.
- Recommendation, ranking or personalisation systems.
- Marketplace or e-commerce experience.
- Causal inference or uplift modelling.
- Embeddings and semantic search.
- A feature store or real-time model serving.
- Fine-tuning open models.
Skills
Interested in this role?
You'll apply directly on UNiDAYS's careers site.