Software · Data · AI · Zürich
Jung Intelligent Systems builds software that works with data: apps and dashboards, automated data flows across many sources, and machine learning up to deep learning, reinforcement learning and robotics. From the first prototype to code running in production.
A neural network learning to separate two interleaved spirals.
Services
Smart data flows that bring several sources together and run on their own. New features derived from the combined data, and dashboards that turn it into knowledge you can act on.
Product development from first prototype to store release: interface, backend, data model and the integrations around it.
Neural networks for images, text and time series, trained on your data and built into your product. With evaluation, monitoring and a plan for retraining.
Agents that learn by trial and error, safely in simulation first. For control, scheduling and other sequential decisions where fixed rules fall short.
Perception, motion planning and control for robots and automated machines, from kinematics to learned policies running on real hardware.
A second opinion on architecture, model choice or build versus buy. You get a written recommendation your team can act on.
Process
We define the problem, look at the data you have and agree on what success means in numbers.
A working version on real data, early enough that changing direction is still cheap.
Production code integrated into your systems, with tests, logging and monitoring.
Documentation and a clean handover to your team. Ongoing support by the hour if you want it.
Recent work
Consumer app
Native iOS and Android app for a Swiss company, built end to end: apps, backend and hosting in a Swiss data centre. It brings flights, places, plans and packing lists from many sources into one clear overview of each trip, works offline and stays up to date in the background. AI features turn bookings and boarding passes into structured trips and suggest what to pack.
native iOS & Androidoffline-firstbackend & Swiss hostingAI featuresdata integration
Applied ML
A measuring device records optical spectra of liquids and soft foods to check their quality and detect adulteration. Up to three chained models evaluate each scan: a classifier screens the sample, a second model identifies the cause of a deviation, and a regression estimates its extent. Several model families were benchmarked with validation across measurement days so results hold up under day-to-day drift, and every model goes through training, evaluation and activation with one-click rollback. Around this sits a web and desktop app with device connection and calibration, offline-first storage with cloud sync, over-the-air updates and guided workflows for operators without ML knowledge.
spectroscopyfood qualitychained ML modelshardware integrationoffline-first & sync
Client names and references on request.
Working together
Every project starts with a short call. After that you get a written scope and a price before any work begins.
Contact
Write a few lines about what you want to build and what data you have. You'll get an answer within two working days.