FIG. 1.0
Hosting services
Managed hosting on your servers, or on bare-metal and GPU machines we provision and run for you. Everything kept reproducible with Ansible , containerised or not.
00 · Data & Infrastructure Engineering
Bespoke software, data and monitoring for your servers.
Hedron is a small studio for collecting, managing and monitoring your data, services and servers. We design the system, build it, and keep it running, on your infrastructure or on ours.
01 · About
A small studio, deliberately
Small is a choice, not a stage. You talk to the people writing the code, decisions happen in one conversation rather than three, and the engineer who designed your pipeline is the one still maintaining it a year later.
We work from Sarajevo, the capital of Bosnia & Herzegovina, and we prefer systems that are boring to operate: reproducible, well instrumented, and dull at three in the morning.
Currently Building data pipelines and low-latency monitoring for clients across the world in Clojure, Java and C++.
02 · What we do
Four disciplines
From a single server to a continuously running data pipeline. We take the work from design through to the alert that wakes someone up.
FIG. 2.0
Bespoke, low-latency monitoring
Anyone can stand up Prometheus and Grafana. We build the case that does not fit it: stream processing over your event flow with Riemann , alerting in seconds rather than scrape intervals, reading from your sources and writing to whichever store you already run, be it PostgreSQL, VictoriaMetrics, InfluxDB, OpenSearch or Elasticsearch.
FIG. 3.0
Software crafting
Long-lived backend systems, and the integration work that sits between services nobody wants to touch. We take on existing JVM codebases as readily as new ones, from low-level C and C++ up to Clojure, Java and Scheme.
FIG. 4.0
Data pipelines, extraction and RAG
Scalable scraping and transformation, turning unstructured text into typed data. Retrieval and LLM workloads run on the JVM with LangChain4j , alongside the pipeline, over the database and data you already operate. Open models hosted on private GPUs when data cannot leave your infrastructure.
03 · Technology
Tech we enjoy working with
We are not religious about tools. These are simply the ones we reach for first. Some of them are deliberately unfashionable: a Lisp and a stream processor will beat a popular default on the problems we tend to get called about. Where you already have a stack, we work with it rather than replacing it.
04 · Clients
Some of the people we work with
05 · Contact
Let's talk about your project
You have an idea, proposal or you simply want to say "Hi"? We will be happy to get in touch and will get back to you.