ml engineer

The company builds interactive, AI-powered video experiences.

• Build and scale the end-to-end RecSys pipeline with two-stage candidate retrieval and ranking for real traffic;
• Implement low-latency, real-time serving models and feature freshness mechanisms for video and infinite-scroll feeds;
• Own ranking and retrieval models using PyTorch, TensorFlow, CatBoost, XGBoost, and sequential algorithms such as SASRec and BERT4Rec;
• Productionize models end-to-end with Python and FastAPI independently;
• Work with product metrics including CTR, watch-time, retention, and NDCG to drive business outcomes;
• Drive architectural direction, set engineering standards, and mentor RecSys team members.

• Direct hands-on RecSys experience with video streaming, social feeds, shorts, or UGC content feeds; general and e-commerce catalog RecSys experience does not apply;
• Strong experience with two-stage architecture: Candidate Retrieval → Ranking;
• Experience with real-time and online serving;
• Solid Python fundamentals, including data structures and algorithms;
• Advanced SQL;
• Production experience with PyTorch or TensorFlow and gradient boosting using CatBoost, XGBoost, or LightGBM;
• Independent backend and productionization skills with Python, FastAPI, or an equivalent framework;
• Experience with workflow orchestration and experiment tracking using Airflow, MLflow, Kubeflow, Dagster, or W&B;
• Strong product mindset and ability to connect ML models to user impact;
• Nice to have: depth with sequential or session-based recommenders such as SASRec and BERT4Rec, strong ML research or model development depth, a proven track record in A/B testing and ranking metric ownership, prior technical leadership or mentorship experience required for Lead level.

• Contract (B2B);
• Flexible hours;
• Direct impact on a fast-evolving product in high-load video technology;
• Supportive, highly skilled engineering team that values clarity and ownership;
• Modern tech stack and access to all necessary infrastructure.