Hyper AdTech is one of the largest programmatic platforms in the Russian market, focused on mobile advertising and in-app inventory.
We operate in open RTB, are integrated with the largest mobile inventory providers, and see a significant portion of the Runet audience – up to 90% of users. This gives us a large stream of bidstream data, advertising events, signals on applications, devices, geo, time, creatives, impressions, clicks, and conversions.
We are now strengthening our ML division and looking for a Senior Data Scientist / ML Engineer who can build models and algorithms that influence the effectiveness of advertising campaigns, the quality of traffic acquisition, and the platform's economics.
Responsibilities
Role Tasks
We want to develop performance in mobile programmatic advertising: more accurately select impressions, predict the probability of target actions, manage bids, improve traffic quality, and enhance campaign economics.
This is a role at the intersection of ML, large event data, advertising auctions, and business metrics.
What you will do
- Build ML models for predicting advertising events: clicks, conversions, installations, leads, purchases, and other target actions.
- Work with large streams of event data: bidstream, impressions, clicks, conversions, campaign data, applications, devices, geo, time, creatives, and traffic sources.
- Develop models and algorithms for performance campaigns: CTR/CVR prediction, traffic quality scoring, bidding, budget optimization, pacing, look-alike/segment scoring.
- Search for features that help better estimate the probability of a target action and the quality of an advertising interaction.
- Perform feature extraction and feature engineering on large log datasets.
- Test hypotheses in offline evaluation and bring successful solutions to online tests.
- Develop and track model and campaign quality metrics: CPA, CPL, ROAS, CTR, CVR, margin, win rate, calibration, uplift, traffic quality.
- Analyze reasons for performance degradation: model, data, inventory, segments, creatives, bids, attribution, or traffic quality.
- Together with the engineering team, deploy models to production: pipelines, APIs, monitoring, retraining, degradation control.
- Identify anomalies, low-quality traffic, and patterns that degrade advertiser results.
Requirements
What is important to us
- Strong Python skills.
- Experience building ML models on tabular, event, or time-series data.
- Solid knowledge of classical ML: classification, regression, ranking, probabilistic models, calibration.
- Experience with pandas, numpy, scikit-learn, and modern ML libraries.
- Experience with CatBoost / LightGBM / XGBoost or similar tools.
- Understanding of mathematical statistics, hypothesis testing, sample quality, bias, leakage, confidence intervals.
- Experience building production ML or closely working with an engineering team to deploy models to production.
- Ability to work with large logs, dirty data, and unstable sources.
- Skill in deep problem analysis: not just "a metric dropped," but why it dropped and what to do about it.
- Business thinking: it's important to understand how an ML solution affects campaign effectiveness, cost per target action, profitability, and acquisition quality.
Will be a plus
- Experience in AdTech, MarTech, DSP, SSP, RTB, mobile in-app, or performance marketing.
- Experience with CTR/CVR prediction, bidding models, recommender systems, ranking, marketplace/e-commerce ML, risk scoring, or antifraud.
- Experience with ClickHouse, Spark, Airflow, Kafka, or other tools for big data and streaming pipelines.
- Understanding of attribution, post-click/post-view logic, traffic quality, and fraud.
- Experience with online experiments, A/B tests, holdout groups, and evaluating the business impact of models.
- Experience with APIs and production pipelines.
Why this might be interesting
- Lots of data: a large stream of advertising events and access to a significant portion of the Runet mobile audience.
- Real production environment: models must impact advertising campaigns, traffic acquisition, and platform economics.
- Complex subject area: RTB auction, in-app inventory, performance, attribution, traffic quality, bidding, and budget optimization.
- Opportunity to build an ML division where results are measured not only by offline metrics but also by money: CPA, CPL, ROAS, margin, campaign quality.
Conditions
- Official employment under the Labor Code of the Russian Federation in an accredited IT company.
- Office in Moscow on Patriarch's Ponds, a 5-minute walk from Mayakovskaya metro station.
- Flexible start of the workday from 09:00 to 11:00.
- Hybrid format: one remote day per week.
- Salary is discussed individually with the successful candidate.
- Voluntary medical insurance with dental coverage after the probationary period.
- Deep immersion into AdTech, programmatic, and performance advertising.
Selection Stages
- Online Interview with HR and Head of Department Discuss experience, motivation, work cases, and Hyper tasks.
- Technical Interview Review ML cases, working with event data, quality metrics, experiments, SQL.
- Meeting with the Team in the Office Discuss work format, expectations, tasks, and potential offer.
