Software our clients rely on
Backend systems, data pipelines, chemometrics and computer vision, designed and built by senior engineers and kept running long after delivery.
Backend and integration
APIs, services and storage on AWS, with authentication, background processing and deployment pipelines, and the glue between existing systems, third-party services and hardware, all delivered with the monitoring and runbooks needed to operate them.
Recent work has included single-table DynamoDB designs, multi-region infrastructure as code, zero-downtime deploys, relay links to devices behind customer firewalls, and numeric code kept bit-identical across Go, C++ and Python with conformance tests.
Learn moreData engineering
Ingest, validation, transformation and reporting for measurement and production data, built as pipelines that are tested, documented and cheap to run, with the analysis on top when it is needed.
Typical pieces are batch and streaming on one schema, idempotent reprocessing of years of history, data quality gates, checks of measurement data against physical constraints, and uncertainty carried through to the reported number.
Learn moreChemometrics
Multivariate statistics on spectroscopic and process data, from PCA, PLS and OPLS regression to PLS-DA and SIMCA classification, with validation designed to hold up outside the calibration set.
Depending on the problem that can mean MCR-ALS or PARAFAC on multiway data, EMSC and scatter correction, calibration transfer between instruments, multivariate process control with T² and Q charts, or designed experiments so the model learns the effect rather than the day.
Learn moreComputer vision and machine learning
Detection, classification and measurement from images and video: model training and validation, deployment on edge hardware or in the cloud, and the data pipelines and monitoring around the models.
Examples from recent projects are transformer-based multi-object tracking, stereo and mirror-based 3D reconstruction with sub-pixel calibration, rolling-shutter compensation, detectors trained on physically based simulation, and TensorRT deployment on edge GPUs.
Learn moreAbout us
bitsort is an independent software company based in Denmark and working with clients across Europe.
The company builds and operates software for engineering and industrial businesses: backend systems, data pipelines and the applications on top, with a background in chemometrics, multivariate statistics, computer vision and machine learning. Projects are staffed by senior engineers, run in small teams, and delivered with the operations work included, so the systems keep working after the project ends.
The language and platform follow the problem and the client's existing systems, not the other way round. Recent projects have been delivered in Go, Python, C++ and TypeScript on AWS; other stacks are taken on where that is the right fit.
Company
bitsort ApS, an independent software and data science company
Based in
Denmark, working with clients across Europe
Contact
info@bitsort.io
Stay in touch
Write to us about a project, a pilot or a second opinion on an architecture.
- info@bitsort.io
- Denmark