Here are a few real‑world projects I’ve worked on as an AI consultant, data scientist and team lead. These are not academic demos, but production cases with clear business context.
LTV model for a health & wellness app
Goal. Predict customer lifetime value (LTV) and help the team optimize marketing and product decisions.
What we did.
– Collected and cleaned data on subscriptions, payments and user behaviour.
– Built an LTV model and a data pipeline for regular reporting.
– Helped the team move from intuition‑based decisions to LTV‑driven planning.
Job text classification (ROCAUC ~0.979)
Goal. Automatically assign categories to job postings for an aggregator, based on description text.
What we did.
– Designed a text classification model for job descriptions.
– Achieved average ROCAUC around 0.979 on evaluation data.
– Integrated the model into the product pipeline, reduced manual work and improved category quality.
Geo‑outlier detection (F1 ~0.95)
Goal. Detect anomalies and outliers in geospatial data (wrong coordinates, strange points, technical noise).
What we did.
– Built a geo‑outlier detection model.
– Achieved F1 score around 0.95 on test data.
– Improved overall data quality and reliability of downstream analytics.
Medical image analysis
Goal. Assist doctors and radiologists in diagnosing diseases from X‑ray, CT and mammography images.
What we did.
– Developed computer vision models for classification and segmentation of medical images.
– Designed a microservice architecture and integration with PACS systems.
– Deployed the solution into clinical workflows, taking reliability and explainability seriously.
Big Data & Data Science department at VEON Kazakhstan (Beeline)
Goal. Increase company profit through data and AI, and build a scalable setup for Big Data & DS.
What we did.
– Led the Big Data & Data Science department with 35+ people and 7 streams (fintech, devices, ads & GEO, computer vision, NLP, internal products, CVM).
– Launched dozens of models and solutions: scoring, risk models, recommendation systems, geo‑analytics, CV/NLP services.
– Standardized processes around data and ML: Code Review, CI/CD, internal tools, regular meetups and team development.
– Achieved a significant profit increase within a year.
If you’d like to discuss a similar project or build your own AI/DS case, you can book a consultation here: https://consultation.alimbekov.com