AI Hub
Real-time ranking. 50,000 ML models. Autonomous AI agents.
If these aren't just buzzwords to you — welcome home.
Preview our contributions
View our conference talks, presentations and articles
Predicting the "Buy It Again" Moment: Engineering Recurring Purchases
Using machine learning models to predict the time to next purchase and power targeted "buy it again" — Data Since Summit 2022
Research papers
Explore our research papers shaping the AI landscape
About Us
The Allegro AI Hub is a high-bandwidth ecosystem where silos between research, engineering, and product vanish. As one of Europe's largest AI teams, we unite Data Scientists, ML Researchers, Analysts, and Engineers to tackle high-dimensional challenges that lack "off-the-shelf" solutions.
From Multi-Objective Ranking across 300M offers to Personalized Recommendations amidst extreme cardinality, we solve what standard tools cannot. Our strength is our interdisciplinarity: Analysts decode complex data signals, Scientists architect probabilistic frameworks, and Engineers build the low-latency systems that power them. We value the intellectual stamina needed to master Non-Stationary Data and Complex Measurement Loops in an environment where every deployment impacts 20M customers and hundreds of millions of transactions.
Want to see the scale of the challenge?
Below is a glimpse of the problems you will crack once you join us.
Shopper AI
Transforming marketing and own media effectiveness by deploying end-to-end AI and GenAI solutions that personalize the customer journey, automate content creation, and optimize commercial profitability.
Scope
- Model user interests and purchase intent using behavioral embeddings and multi-signal classification (e.g. customer hobby detection)
- Personalize experience and engage customers using State of the Art models (e.g. Graph Neural Networks, Transformers)
- Automate multi-step marketing and content workflows using LLM-powered autonomous agents with reasoning and tool-use capabilities
- Maximize ROI on paid channels while enforcing content standards using classification and ranking models
- Generate and optimize product and marketing content using LLMs with domain-specific fine-tuning (e.g. creative copy writing)
Recommendations
Builds intelligent, large-scale ML systems and deep learning models that connect millions of daily users with relevant offers through a blend of academic research and engineering.
Scope
- State-of-the-art deep learning models
- Representation learning
- Information retrieval and ranking
- Personalization
- User behavior prediction
Learning to Rank
Develops machine learning models for search pipelines, focusing on neural text-based search and relevance to serve millions of daily queries.
Scope
- Ranking solutions for search pipelines
- Neural text-based search and relevance
- Reranking techniques
- Feature interaction architectures
- Personalization
Predictive Commerce
We provide machine-learning based solutions for improving efficiency of managing the platform on the company level, as well as boosting the efficiency of commercial actions.
Scope
- Forecasting of key metrics (eg GMV, Page Views) on multiple levels
- ML-based Pricing (managing discounts)
- Boosting efficiency of deals-related landing pages
Predictive Operations
Enabling the area of Operations, Customer Experience and Delivery Experience success by harnessing the power of Artificial Intelligence. Our objective is to serve as strategic ML partners, continuously adapting our focus to develop bespoke solutions that meet the specific challenges and ambitions of our clients.
Scope
- Predictive Analytics: Forecasting critical business metrics to inform strategic planning
- Customer Intelligence: Uncovering patterns in client behavior to enhance personalization and engagement
- Supply Chain Optimization: Utilizing geospatial and network analysis for strategic localization
- Operational Excellence: Identifying and implementing data-driven improvements to core business processes
- Data Augmentation: Systematically enriching enterprise datasets to unlock deeper insights
Experimentation & Measurement
Our mission is to be the definitive authority on incrementality at Allegro. We aspire to empower every team, regardless of their project or service, with the methodologies and insights needed to drive and measure true incremental value.
Scope
- Test Drive: The incremental value of most projects can be effectively quantified through A/B testing
- CATE Simulation: regression based approach to comparison of two groups
- Time-to-Event Data-Driven Attribution: regression model and event impact time decay
- Instrumental Variable: solving endogeneity issue, only for LATE
Computer Vision
Elevates the user experience by leveraging image processing and multimodal models that integrate visual data with product titles, descriptions, and attributes.
Scope
- Visual Search and image representation learning
- Robust image classification models
- Multimodal data integration
- Semantic search
- Advertising and marketing solutions
- Product catalog quality enhancement
Join Our Team
We're looking for talented individuals who are passionate about innovation across Data Science, ML Research, Product Analysis and Engineering. Help us shape the future of AI-powered e-commerce.