Košice, Slovakia · since 2016 ceo@itrslovakia.com
ITR Slovakias.r.o. · Košice Arrange a consultation

Data analytics · Machine learning · AI

We turn data into decisions that cut risk and raise profit

We design and deploy scoring, predictive and AI solutions for companies that want a result in production — not just a prototype. Finance, media, retail and advertising.

Arrange a consultation See the results
Scoring modelschematic illustration
  • 2016founded
  • 6areas of analytics
  • 5–15years of specialist experience
  • 15–30%profitability gain in a scoring project

Company

An analytics and software company

ITR Slovakia s.r.o. has been operating since 2016. We work on data analytics, machine learning and custom software development for financial institutions, media, retail and advertising agencies.

We run a project from the data audit and model prototype through to production deployment and further development. Our specialists have 5 to 15 years of experience and work in small teams directly with the client.

Arrange a consultation

Services

What we do for clients

Six areas where we have both ready models and production experience.

AI solutions

We deploy large language models into specific business processes and connect them to your databases and applications.

  • AI assistants and chatbots
  • AI agents and process automation
  • RAG — answers over company data
  • Document and email processing
  • ChatGPT, Claude and Grok integration

Finance

Scoring and risk models on credit, transaction and CRM data — decisions on applications in seconds and a more accurate view of problem borrowers.

  • Customer analytics and segmentation
  • Financial modeling and scoring
  • Risk identification and fraud detection
  • Credit analytics and automated decisions
  • Marketing analytics

Media and social networks

Content personalization and user retention based on browsing history, activity and clickstream; measurement of advertising effectiveness.

  • Content personalization
  • Customer retention
  • Recommender systems
  • Predictive analytics
  • Sentiment analysis

Retail

A 360° customer profile from purchase history and online activity; optimization of assortment, pricing and stock planning.

  • Personalized offers
  • Store design and ergonomics
  • Inventory and demand management
  • Assortment optimization
  • Dynamic pricing

Advertising

Audience modeling, multichannel campaign attribution and optimization of budgets and ad placement.

  • Customer assessment (client 360°)
  • Best offer at the right time
  • Risk management
  • Sales optimization
  • Sentiment analysis

Computer vision

Recognition of documents and faces from photos and video streams — for client onboarding and identity verification.

  • Document recognition: passport, ID card, driving licence, vehicle registration
  • Face recognition from video and photo
  • Gender, age and expression detection
  • Image quality checks
  • AWS Rekognition integration

Case study

Trustworthy client model for a loan portfolio

The model estimates the probability of loan repayment and assigns every application to one of three categories. Most applications are therefore assessed without manual review.

Client
a financial institution; the name is not disclosed (NDA)
Problem
manual assessment of loan applications was slow and imprecise
Data
the client's credit, transaction and CRM data
Solution
a repayment probability model splitting applications into three decision bands
Deployment
the model is connected to the client's decision process
  • 15–30%higher portfolio profitability
  • 2×fewer underwriters
  • < 1 minto process one application

We do not disclose the client or the size of its portfolio (NDA). The size of the effect depends on the client's current business process.

  1. Rejectionlow probability of repayment
  2. Additional verificationborderline band, decided by an underwriter
  3. Automatic approvalhigh probability of repayment
Probability of repayment: lowhigh

How we work

How a project runs

  1. 01

    Data audit

    We review the available data, sources and processes, and define what is measurable and where the gain is largest.

  2. 02

    Model prototype

    We build the model on historical data and show the expected effect on your own numbers.

  3. 03

    Deployment

    We connect the model to your systems — API, CRM, databases — and tune it in production.

  4. 04

    Support and development

    We monitor model quality, retrain it on new data and extend its functionality.

Technology

What we work with

Generative AI

  • ChatGPT (OpenAI)
  • Claude (Anthropic)
  • Grok (xAI)
  • RAG

Machine learning & data science

  • Python
  • scikit-learn
  • XGBoost
  • Computer vision

Cloud & infrastructure

  • PostgreSQL
  • ClickHouse
  • AWS
  • Docker
Show the full technology stack
Generative AI
ChatGPT (OpenAI) · Claude (Anthropic) · Grok (xAI) · OpenAI API · Anthropic API · xAI API · RAG · embeddings · function calling · AI agents
Machine learning & data science
Python · R · SAS · scikit-learn · XGBoost · Keras · pandas · NumPy · SciPy · NLP · Computer vision · Feature engineering
Models
Scoring · Credit risk · Fraud detection · Recommenders · Segmentation · Classification · Regression · Time series
Development & integrations
PHP · Laravel · Symfony · Flask · Node.js · React · REST API · Microservices
Databases
ClickHouse · PostgreSQL · MySQL · MS SQL Server · MongoDB · Elasticsearch · Redis · Sybase IQ · SQL / T-SQL / PL-SQL
Cloud & infrastructure
AWS (Lambda, S3, Kinesis, DynamoDB, Rekognition) · Azure · Docker · Kafka · RabbitMQ · Jenkins · ELK · Nginx

Specialist experience: 5 – 15 years.

Contact

Tell us about your project

Write to us about the data you work with and what you need to achieve. We will reply with a proposal for the next step.

What happens next

  1. You describe the data you work with and the result you need.
  2. We reply by email — with questions or a proposed next step.
  3. On a call (about 30 minutes) we go through scope, data and a possible approach.