Optimization consultancy, Melbourne
Better decisions, by design
Founded by three MiniZinc core developers. We build and improve optimization models for scheduling, timetabling, routing, and planning, and we support organizations that already run MiniZinc.
Problems we help with
Decisions with too many options to check by hand, and rules you can't break. If your organization faces decisions like these, we can model them, solve them, and build them into your systems.
Scheduling and rostering
Shifts, jobs, and machines assigned so that hard rules hold and the costs you care about go down.
Timetabling
Classes, exams, or events placed in rooms and time slots without clashes.
Logistics and routing
Vehicles, routes, and deliveries planned against capacities and time windows.
Planning and resource allocation
Budgets, people, and equipment allocated across competing demands, including configuration and design problems.
The MiniZinc team built a donation-allocation model for UNHCR that delivered
US$400 million more in allocationsthan the spreadsheet method it replaced
Re-plan a factory in your browser
A rush order arrives, a machine goes down, a deadline moves. Change the factory below and the solver finds a new best schedule. Rosters, timetables, and delivery routes work the same way. This is real MiniZinc, running on your machine.
The solver first minimizes hours late (rush jobs count three times), then the finish time. Change any setting and the solver re-plans. Click a machine row (or M1–M5) to add downtime, up to 6; click a block to remove it.The solver first minimizes hours late (rush jobs count three times), then the finish time. Change any setting and the solver re-plans. Tap M1–M5 to add downtime, up to 6; tap a block to remove it.
Per-job settings: rush order and earliest start (h)
- Job 1 late +2
- Job 2 due 64
- Job 3 due 79
- Job 4 due 38
- Job 5 due 78
- Job 6 due 103
- Job 7 due 63
- Job 8 due 91
Have a schedule like this?
Discuss your problemThe team
We know the language, the compiler, and the solvers from the inside, and we bring in other optimization specialists when a project needs them.
Dr Jip J. Dekker
Jip specializes in making models solve faster, from how MiniZinc translates a model to how the solver searches. He works on the MiniZinc compiler, wrote MiniZinc Python, and leads development of Huub, an open-source solver. He holds a PhD from Monash University (2021) on the architecture of constraint modelling languages.
Jason Nguyen
Jason turns optimization models into working software. He develops and maintains the MiniZinc tool chain, including the compiler and IDE, and wrote MiniZinc JS, the WebAssembly build behind the live demo on this page. He also runs the MiniZinc Challenge, the annual solver competition.
Dr Guido Tack
Guido helps clients bring optimization into their operations. He leads MiniZinc development at Monash University and co-develops Gecode. His projects with larger organizations include donation allocation for UNHCR, metering design for intelligent water networks, and energy pricing with CSIRO Data61.
Built on open source
We work on open-source tools that anyone can use. MiniZinc, developed by our team since 2007, is the state of the art in constraint modelling and has been downloaded hundreds of thousands of times. Huub, the Lazy Clause Generation solver Jip leads, is usable from MiniZinc or as a Rust library. Gecode is a constraint programming toolkit Guido co-develops. MiniZinc and Huub are released under the MPL-2.0 licence, Gecode under the MIT licence.
We also work with commercial and open-source MIP and other solvers, including Gurobi, CPLEX, HiGHS, and OR-Tools CP-SAT.
Consultancy
For teams starting out with optimization, or smaller organizations that need expert help on a specific project. We work on your existing optimization application, whatever solver it uses, or build one with you.
Discuss a projectRunning MiniZinc in production? See the deciZns Partnership →
Model review and tuning
We check your model for correctness and solving speed, then pick and tune the solvers that suit it.
New optimization applications
We build the whole thing: the model, the data pipeline, and the integration into your systems.
Training
Workshops and courses on modelling with MiniZinc and on constraint programming.
Solver and MiniZinc work
Custom solver features, new global constraints, compiler changes, and integration work.
How an engagement works
- 1Talk. Email us about the problem. We arrange a call to understand what you need.
- 2Scope. We agree on what we will deliver and how we will work together.
- 3Deliver. We do the work with your team and hand over models and code you can run and maintain.
deciZns Partnership
For organizations that rely on MiniZinc in their operations. You get a team that knows the internals, available through the year, so your models keep working and stay current as MiniZinc and the solvers change. You also get a say in what we build next. One yearly contract, sized to your organization.
Ask about a partnershipWhat's included
Reserved hours
A set number of our hours each year, for support or for MiniZinc and solver work you choose with us.
MiniZinc updates tailored to you
A quarterly report on what changed in MiniZinc and its infrastructure, and what it means for your models.
Priority for your feedback
Your feature requests and problem reports get priority consideration when we plan MiniZinc and solver development.
Open source and confidentiality
General improvements funded this way go into the public open-source MiniZinc and solver releases. Your own code and data stay confidential.
Contact
What are you trying to decide?
A few lines is enough to start. We work with clients worldwide, remotely.
If it helps, mention the tools you use now and whether you're after a project or a partnership.