Azurro

Azurro is a Polish software development company founded in 2010, based in Warsaw. They specialize in data processing pipelines—data acquisition, storage, search, and analytics—with particular strength in big data and machine learning. The team has worked extensively with companies in the USA and Europe across e-commerce, medical information, patent analysis, and maritime industries.

The company's core work spans system architecture design, cloud deployment (especially AWS and Google Cloud), cluster management, and web development. They're ranked 4th globally on TopCoder for programming competitions. All developers hold university degrees, and the majority speak English fluently.

Services and capabilities

Azurro's technical offerings center on data pipelines and large language models. On the data side, they handle ETL processes, microservices architecture, custom query engines for analytics, and complex deduplication at scale. They process gigabytes of data daily for clients, pulling from APIs, streaming platforms, social media, and custom datasources.

For large language models, they offer consultation on LLM strategy, system design around language models, fine-tuning and customization, integration with existing systems, and cloud or on-premises deployment. Chris Ociepa, co-founder, supervises training and fine-tuning for Bielik, Poland's first open language model. They've built speech-to-text systems using Microsoft Custom Speech, and systems for analyzing and indexing audio and video materials (their recent Matena Analyzer product).

They work with various LLMs depending on client needs and offer RAG (Retrieval-Augmented Generation) architecture implementations. Post-implementation support includes updates, troubleshooting, system improvements, and retraining—addressing the lack of official support around open-source solutions.

Web development uses Django, hosted on AWS (ECS, S3, RDS, ElasticSearch, DynamoDB, ECR) or Google Cloud Platform. Machine learning components employ SVM classifiers, Named-Entity Recognition, and AWS Comprehend for NLP tasks. Testing is core: they emphasize automated test suites, Continuous Integration, and Test-Driven Development.

Notable work

They built an advanced maritime data processing system handling Automatic Identification System (AIS) data—vessels, tracking, port entry/exit detection. The system processes both historical and real-time data with high stability; it ran three years without major issues using microservices and monitoring. Client Simon Rathbone (CTO, MariTrace Ltd) noted the system "exceeded our expectations" and provided "critical insights and a robust data processing framework."

For a US patent prosecution firm, they created a system analyzing large patent decision datasets. Machine Learning and Natural Language Processing extract information from decision texts. A Django web app with multiple search filters, email notifications, and daily document downloads handles the interface. Data lives in MongoDB and Cloud SQL on Google Cloud.

For a US medical information company, they built a system searching news sources for information about specific doctors. Separate microservices process data from different sources. An SVM classifier filters relevance; Named-Entity Recognition identifies doctors by name. Hosted on AWS.

Another medical client received a system processing scientific articles from multiple datasources. The team handled the entire pipeline—fetching, processing, storage, and presentation in a web app. They unified formats across different APIs, deduped documents across websites, and generated KPIs using AWS Comprehend and Comprehend Medical. Data stored in S3, RDS, DynamoDB, ElasticSearch.

A US Big Data analytics firm gets ongoing ETL work. Azurro gathers data from Twitter, YouTube, streaming and billing platforms. They built a custom query engine for aggregation and metrics, handling gigabytes daily.

How they work

Azurro follows Agile and builds in "quick and small functionality increments," giving clients control over the process. They communicate frequently via multiple channels, update status regularly, and ask for clarification when needed. Clients can request improvements and modifications throughout the project lifecycle.

They operate as either self-driven teams or embedded in client organizations, depending on needs. They enjoy POC and experimental work. The time zone between Warsaw and Western Europe, the US East Coast, or Asia is manageable for their communication schedule. They're EU-based and comply with data protection security standards.

No specific pricing model is mentioned in their public material.

Team and credentials

Founded in 2010. Chris Ociepa is co-founder and supervises LLM training work. All developers hold university degrees from Polish universities. Ranked 4th globally by TopCoder in programming competitions. Developers speak English fluently. The company complies with EU data protection standards (Poland is an EU member). No team size, certifications (ISO 27001, SOC 2, etc.), or G2/Clutch ratings are mentioned in the source.