Applied ML Engineer

Job Summary

We are looking for an Applied ML Engineer with strong experience in recommender systems to build the brain of our in-game commerce store. This role involves creating a recommendation engine that decides which products, coupons, and rewards to surface to players at the right moment. The ideal candidate will bring their expertise in collaborative filtering, embedding-based retrieval, and evaluation strategies to enhance our product offerings.

Responsibilities

• Own the collaborative filtering model, starting with Gorse and potentially moving to a custom stack.
• Build product embeddings using techniques like product2vec and Faiss or Approximate Nearest Neighbors (ANN) for the catalogue.
• Evolve cohort assignment from rules-based systems to machine learning-driven approaches.
• Develop the offline evaluation framework focusing on precision@k, NDCG (Normalized Discounted Cumulative Gain), conversion rate, diversity, and coverage.
• Bridge offline models to online serving, including model serving infrastructure and a weekly refresh pipeline.
• Calibrate ranking weights against key business outcomes, such as Click-Through Rate (CTR), Gross Merchandise Value (GMV), margin, and repeat redemption rates.
• Collaborate closely with the Data Engineer on event pipeline and feature store integration, as well as with the Backend Engineer on the ranking API and Redis serving layer.
• Transform sparse and noisy in-game event data into reliable signals for model improvements.
• Act as the internal owner of recommendation quality to ensure the model effectively impacts outcomes rather than just performing well on offline metrics.

Qualifications

• Experience: 3–5 years of machine learning engineering experience, specifically with recommender systems.
• Technical Skills:
  • Proficient in Python with frameworks like PyTorch/JAX, and libraries such as scikit-learn and NumPy.
  • Hands-on experience with collaborative filtering, including sparse matrices and cold start problems.
  • Experience in embedding-based retrieval using tools such as Faiss, ScaNN or equivalents.
  • Strong understanding of recommendation evaluation metrics, focusing beyond accuracy to include diversity, coverage, and business outcomes.
  • Ability to work comfortably with sparse and noisy datasets.
  • Proven experience of taking models from offline development (notebooks) to online production serving.
  • Strong communication skills and structured problem-solving capabilities.

Preferred Skills

• Experience in building voucher, coupon, or deal recommendation systems.
• Background in gaming, mobile, or consumer engagement products.
• Familiarity with frameworks such as Gorse or LightFM.
• Experience with contextual bandits or online learning methodologies.
• Knowledge of feature store patterns.
• Previous experience in startup environments or roles requiring ownership in fast-paced settings.

Experience

• Minimum of 3–5 years of related experience, particularly in machine learning engineering with an emphasis on recommender systems.