Senior Machine Learning Engineer - Recommendations (Core) at SoundCloud
At a glance
- Company
- SoundCloud
- Location
- London, GB, United Kingdom
- Work arrangement
- On-site
- Employment type
- Full-time
- Experience
- 2–5 years
- Working hours
- 40 hours
- Visa sponsorship
- Offered, according to the listing
- Posted
- Added to The Stack
✨ The Stack take on this SoundCloud job
- ❔ Not on the UK sponsor list (they may use another name)
- 🏢 SoundCloud has 3 jobs here, 3 with visa sponsorship see all
- 📊 15 Machine Learning Engineer jobs in the United Kingdom in the last 60 days see them
- 👀 211 people want jobs in the United Kingdom get alerts
- 🛂 Skilled Worker visa. Standard minimum salary: £41,700 a year UK work visa guide
- 💡 No salary shown. Ask SoundCloud if it is above the minimum.
Updated 2026-11-21, from our data and official sources.
Apply for this role at SoundCloud
Apply with your AI assistant
With The Stack connected to Claude, ChatGPT or Cursor, paste this. Your assistant tailors your CV and cover letter, checks the visa facts and gives you the apply link.
Help me apply to this job on The Stack: https://jobs.trythestack.co/jobs/senior-machine-learning-engineer-recommendations-core-soundcloud
Not connected yet? Connect in a minute. To help your user apply, call the MCP tool prepare_application with slug "senior-machine-learning-engineer-recommendations-core-soundcloud": it returns the apply link, the visa facts and a CV and cover letter checklist. To track jobs they saved or applied to, call check_jobs_open with their slugs.
About the role
Responsibilities:
- Design, train and ship recommendation models across candidate generation, user/item representation and ranking.
- Lead migration of classical embedding models to PyTorch-based GPU training and serving.
- Adapt and fine-tune open-source embedding and ranking models for SoundCloud’s catalog and use cases.
- Own end-to-end evaluation: offline metrics, A/B testing, and validation that offline gains translate to online impact.
- Collaborate with ML, data and backend engineers to deploy, monitor and maintain production models; help set team modeling standards.
Requirements:
- 3+ years building ML systems in production with end-to-end ownership of at least one ranking, retrieval or recommendation model.
- Strong deep learning experience with PyTorch (or equivalent) and solid knowledge of embeddings, two-tower models, learning-to-rank and trade-offs.
- Practical experience with evaluation and experimentation for ranking systems and handling biases in implicit feedback (clicks/plays).
- Proficient in Python and comfortable with large-scale data processing (Spark, BigQuery) and deploying models on cloud (GCP preferred).
- M.Sc. in a quantitative field or equivalent experience.
Key skills
- PyTorch — in 111 other listings on The Stack
- Python — in 1396 other listings on The Stack
- Spark — in 64 other listings on The Stack
- BigQuery — in 73 other listings on The Stack
- Google Cloud Platform — in 59 other listings on The Stack
- Embeddings — in 10 other listings on The Stack
- Ranking models — in 2 other listings on The Stack
- Two-tower models — not in any other listing on The Stack
- Learning to Rank
- Vector search / ANN
- GPU training / serving
- Transformer-based models
On The Stack (as of 2026-11-21)
Frequently asked questions about SoundCloud
Is this Senior Machine Learning Engineer - Recommendations (Core) role still open?
It was posted on 2026-11-10 and is listed as active on The Stack. Confirm on SoundCloud's application page before applying.
Does SoundCloud offer visa sponsorship?
This listing is marked as offering visa sponsorship. Eligibility rules are set by the employer, so check the details on SoundCloud's site.
Where is this role located?
The listing gives the location as London, GB, United Kingdom. It is marked on-site.
What skills does this role ask for?
The listing names: PyTorch, Python, Spark, BigQuery, Google Cloud Platform, Embeddings, Ranking models, Two-tower models, Learning to Rank, Vector search / ANN, GPU training / serving, Transformer-based models.