ml engineer for personalization and monetization

Описание: Higgsfield AI is a generative AI company focused on AI-powered video creation and next-generation creative tools. Its products serve more than 25 million users worldwide and support content generation for major global brands. Задачи: Build ranking systems for effects, presets, templates, models, prompts, and post-generation recommendations; Solve cold-start personalization for new users with limited signals; Address position bias, popularity feedback loops, exploration versus exploitation, and logged-behavior bias; Optimize ranking for completed, retained, and shared generations rather than clicks; Own incrementality for discounts, vouchers, trials, upgrades, and win-back campaigns; Design and maintain randomized holdouts for uplift modeling and measurement; Apply margin and abuse guardrails to offer optimization; Work with Legal on personalization, pricing, discounting, consumer protection, and consent requirements; Build churn and downgrade prediction for subscribers and repeat-purchase propensity models for credit-pack buyers; Pair propensity models with interventions and experiments to improve retention; Monitor for data leakage in churn and retention models; Create use-case and intent features from prompts, input assets, output assets, models, and parameters; Analyze failed, abandoned, and refunded generations as churn and unmet-demand signals; Own models end to end from problem framing and feature development through training, evaluation, serving, monitoring, and retraining; Ship models behind experiments and use online results to determine what remains in production; Monitor model drift and degradation caused by launches, pricing changes, and seasonality; Track and maintain the business value of each model in monetary terms. Требования: Experience shipping machine learning models serving live user traffic and changing business metrics; Depth in at least two areas: recommender systems, learning-to-rank, uplift and causal ML, churn or propensity modeling, real-time personalization; Strong causal understanding of randomized holdouts, incrementality, Qini and uplift evaluation, selection effects, and before-and-after bias; Strong Python and SQL skills; Experience with gradient boosting, neural ranking, feature pipelines, training-serving skew, latency budgets, and retraining cadence; Product judgment to define the decision and metric before selecting a model; Comfort working with images and video as data, or willingness to learn quickly; Pragmatic approach to shipping heuristics and replacing them when appropriate; Clear written and spoken English at B2+ level; Not suitable for candidates who want to train or fine-tune generative video models, optimize offline metrics without deployment ownership, target users only by conversion propensity, rely on before-and-after lift comparisons, require clean labeled data and an existing feature store, or want predictable 9–5 workdays; Nice to have: Background in recommender systems or ranking for scaled consumer products, growth or monetization ML in subscription, gaming, fintech, or e-commerce, causal inference or uplift systems, applied science with production model ownership. Условия: Competitive base salary in USD, based on experience, skills, and role scope; Equity participation through the company’s stock option program; Relocation support to Almaty for candidates moving from another city or country; Collaborative, fast-paced environment with direct access to experienced leaders; Opportunities for professional growth, ownership, and career development; Company-provided equipment, meals, transportation, and other office benefits; Office work five days per week for the full working day.