Overview
In this role you will develop and deploy cutting-edge ML to personalize Bumble’s recommendations at scale. You’ll own problems end-to-end, from data exploration to model production, collaborating with cross-functional teams to improve user experiences. You’ll apply deep learning frameworks to train and optimize models in production and contribute to experimentation and evaluation pipelines. With a focus on responsible AI, you’ll help ensure fairness and safety in model deployment, driving impact from insight to action.
Responsibilities
- Explore, develop, and deliver ML tech to personalize recommendations at scale
- Own end-to-end problems from data exploration and feature engineering to model training, evaluation, and deployment
- Design, train, and optimize production models using PyTorch or TensorFlow
- Contribute to experimentation frameworks, including A/B testing and offline evaluation
- Maintain and monitor production models; diagnose issues and iterate for reliability
- Deliver high-quality solutions from insight to impact with speed and rigor
- Apply responsible AI practices focusing on fairness, transparency, and safety in model development and deployment
Key requirements
- Around 3 years of hands-on experience shipping ML models in production
- Strong Python programming skills and proficiency with PyTorch or TensorFlow
- Experience in researching or applying ML, especially in recommender systems, ranking or personalization
- Good understanding of MLOps and infrastructure concepts (CI/CD for ML, feature stores, model serving, observability, versioning)
- Familiarity with containerisation and cloud-native environments (Docker, Kubernetes, GCP)
- Familiarity with experimentation methodologies (A/B testing, model evaluation techniques)
- Agile mindset with ability to adapt approaches based on data and priorities
- Growing AI fluency with ability to apply ML techniques and emerging tools (including LLMs) to problems
- agile mindset
- data-driven decision making
- responsible AI awareness
- Python
- PyTorch
- TensorFlow
…
