Services
AI solutions
Use AI that works. Language models, agents, time series and classic machine learning: we pick what fits, tune it to your data, and ship it. Sensitive data? We run it on your own hardware.
Software engineering
Ship software you can trust. Code review, automated tests and continuous integration are our defaults, not extras. Agile delivery with full visibility, and AI-assisted development that makes us faster without lowering the bar.
Data science
Ask better questions of your data. We use statistics, probability models and simulations to turn business questions into answers. We also build the pipelines and reporting underneath, so the numbers stay trustworthy.
Performance and scalability
Make it fast, and keep it fast. Speed, throughput and low latency are features. We profile backends and frontends to find the real bottleneck, then scale hardware, networks and software to meet demand.
Business consulting
Grow on purpose. We help you pick the right opportunities, set goals that stick, and match your data with technology that grows revenue.
Product design
Build less, and make it count. A great app is as much about what you leave out as what you add. We help you test with real users, prototype on paper and on screen, and run UI experiments that show what a feature does for your metrics.
Innovation
Make luck a process. The best ideas don't wait for a lone genius. We help you carve out time to explore, run cheap experiments that can fail safely, and reward the people brave enough to try.
Media and marketing
Spend where it pays. Online ads, content and SEO, tracked against what they earn, so you can cut what doesn't work and spend more on what does.
Capabilities
AI
| Category | Items |
|---|---|
| machine learning | deep learning (neural networks), language models, embeddings, time series, classic ML, model evaluation and tuning |
| models | agents (MCP, harness development), local LLMs (llama.cpp, vLLM), local speech and image models |
| applications | retrieval-augmented generation (RAG), entity extraction (NER), document extraction and OCR, classification, summarization, chatbots, workflow automation |
| hardware | CUDA, ROCm and Vulkan stacks, multi-node clusters for local inference, on‑premises deployments |
| quality | evaluation suites, benchmarks on your own data, regression testing for prompts and models, guardrails, cost and latency tracking |
Engineering
| Category | Items |
|---|---|
| practices | AI-assisted and agentic development, code review, automated testing, fuzzing, end-to-end testing, continuous integration |
| languages and platforms | Dart, Flutter, Python, Java, Go, Rust, TypeScript, C and C++, WebAssembly, Android, iOS |
| infrastructure | Linux, Docker, Terraform, cloud, observability, self-hosting |
| security | threat modeling, secrets management, access control, data privacy and anonymization (GDPR) |
| performance | profiling backends and frontends (sampling profiler, browser tooling), load testing, scaling hardware and networks, high-performance computing |
Data
| Category | Items |
|---|---|
| data science | statistics, probability models, simulations, forecasting, causal inference, clustering, hypothesis testing, A/B testing |
| quality | data validation, deduplication, lineage and provenance, anomaly detection |
| databases | relational and distributed SQL (PostgreSQL, CockroachDB), NoSQL (Cassandra, DynamoDB), analytics (ClickHouse, DuckDB) |
| search | full-text and vector search (Lucene, Solr, Elasticsearch, OpenSearch, pgvector), knowledge graphs |
| relevance | (re)ranking, query understanding, multilingual, geospatial and multimedia search |
| pipelines | streaming and batch processing (Kafka, Spark, Flink) |
| visualization | dashboards, interactive charts, maps, Grafana, Apache Superset |
Investigations
| Category | Items |
|---|---|
| investing | risk modeling, portfolio analytics, backtesting, pricing models |
| compliance | sanctions screening, adverse media screening, AML, KYC, due diligence, conflict of interest analysis |
| fraud | ownership mapping, entity resolution, fraud detection |
Team
Small crew, senior people: we get more done with less ceremony.
Mária Molnár
project manager, QA
(ex-Accenture)
AI agents (on-demand)
engineering, research and testing, around the clock
(self-hosted or cloud APIs)
For bigger jobs we bring in trusted experts and partner companies. Some of them:
Contact
Based in Budapest, working with startups and multinationals, on-site or remote. Tell us what you're building or stuck on, and we'll tell you if we can help.
