SPT | Senior Machine Learning Engineer

Overview
About CBTW
CBTW is a global independent tech consultancy group. With a team of more than 3,000 talented tech specialists on the ground in over 35 cities across Europe, the USA, Asia, Australia and Africa, we serve clients of all sizes. We believe collaboration betters our world. Our unique open ecosystem model allows us to deliver end-to-end tech solutions.
In this role, you will join a team building recommendation and personalization systems for one of our partners, a leading video streaming platform.
Responsibilities
A hybrid ML/data engineering role that owns recommendation and personalization systems end to end. It covers feature pipelines, model training, low-latency serving, monitoring, and iteration. The position sits between data platform engineering (streaming, lakehouse, orchestration) and applied ML (ranking, embeddings, experimentation), with full production ownership after launch.
Core Responsibilities
Own recommendation and personalization models from research through production: training pipelines, offline evaluation, A/B testing, and post-launch performance.
Build and run low-latency serving APIs and feature store pipelines, and design the batch and streaming data pipelines behind them (Kafka/Flink, Spark, Airflow, BigQuery).
Keep everything reliable through monitoring, alerting, and data quality checks, root-cause production issues, and optimize cost and latency. Document architecture and schemas so the wider organization can self-serve.
Qualifications
Experience: 5+ years in data science, ML engineering, or data engineering, with models and pipelines running in production.
Recommendation/ranking: Hands-on experience building and shipping recommendation, ranking, or personalization systems (embeddings, vector search/ANN), including label design, negative sampling, cold start, and avoiding data leakage.
Experimentation: Offline evaluation and A/B test design and analysis, with the judgment to tell real gains from noise and to translate offline results into online impact.
Programming: Strong Python and advanced SQL; pandas, scikit-learn, and TensorFlow or PyTorch.
Data pipelines: Experience with streaming and distributed processing (Apache Kafka, Apache Flink, Apache Spark) and Apache Airflow / Cloud Composer, on GCP (BigQuery, GCS, Dataproc). AWS (S3, Redshift, Glue) migration experience valued.
Serving & engineering: Production low-latency APIs (e.g. FastAPI), with solid software engineering practices: modular code, testing, Git, CI/CD, and Docker. Able to reason about latency, throughput, cost, and failure modes.
Nice to Have
Apache Iceberg or other open table formats; Dataform; GitLab CI; infrastructure as code
Vertex AI and managed feature stores; low-latency serving stores (Redis, Bigtable)
Monitoring and data quality tooling (e.g. Datadog, data catalog and lineage)
LLMs, RAG architectures, and GenAI applications; graph-based similarity or graph ML
Streaming/media domain experience, including video and shorts analytics