Machine Learning Engineer | Payments ML Pipelines and Recommendation Models

Adyen is the Dutch payments platform that processes transactions for companies like Spotify, Uber and eBay, running its own infrastructure end to end rather than reselling someone else's. This Bengaluru role builds and maintains the production machine learning services behind its data products. The work is full lifecycle: developing production ML pipelines, adapting global frameworks to India specific requirements while keeping architectural parity, and designing recommendation and classification models that go directly into merchant facing products. There is a strong performance thread too, finding and fixing memory, latency and throughput bottlenecks in training and inference so models hold up in a high throughput payments environment. At 4+ years with expert Python expected, it sits in the band where an ML engineer has shipped real systems but is not yet expected to have run a team.

Required
- 4+ years of experience in the machine learning domain, with expert level Python skills and deep familiarity with the standard data science toolkit: PyTorch, TensorFlow, XGBoost or LightGBM, Pandas and Scikit-learn.
- Proficiency across the end to end ML lifecycle, from building robust pipelines on big data (Spark, SQL, Trino) through deploying and maintaining models in production using MLOps practices.
- Hands on experience with ML infrastructure and orchestration tools such as Kubernetes, Docker, Airflow and Argo.
- Software and data engineering best practices, applied in partnership with international MLOps and platform teams.

The day to day
- Developing and maintaining full lifecycle production ML pipelines, and acting as a technical anchor adapting global frameworks to India localised requirements.
- Designing and productionising recommendation and classification models integrated directly into merchant facing products.
- Identifying and resolving performance bottlenecks in training and inference across memory, latency and throughput.
- Owning reusable AI components that form the foundation for scaling generative AI applications.
- Partnering with international MLOps and platform departments to adopt internal tooling.

Location and working style
Bengaluru. Adyen states its position without ambiguity: it is an office first company that values in person collaboration and does not offer remote only roles. Take that at face value. If remote is what you need, this is not the posting for you, and the LiveKit and 6sense roles in this same edition are genuinely remote.

Honest fit guidance
The phrase to focus on is production ML services, not model research. Adyen wants an engineer who can own pipelines, deployment and performance, and the requirement list is weighted accordingly: MLOps tooling, Kubernetes, Airflow and Spark sit alongside the modelling libraries. If your experience is training models in notebooks and handing them over, this role will stretch you in the infrastructure direction.

The India localisation thread is worth asking about. The posting describes adapting global frameworks for India specific requirements, which suggests genuine ownership rather than following a specification written elsewhere, but it also implies coordination with teams in other timezones. Ask how that works in practice.

Payments is a demanding domain: high throughput, low tolerance for error and meaningful regulatory weight. That is what makes the engineering interesting.