We are two optimization PhDs building hybrid AI + Operations Research systems for planning, routing, scheduling, pricing, and allocation. AI helps with interfaces, explanations, forecasts, and rule editing; optimization keeps decisions within constraints, testable, and defensible.
Prior engagements include
swissQuant
Quantitative finance · risk-aware portfolios
Italmondo
Logistics · production container loading
ALPIQ
Energy trading · 900 MW pumped-storage
UNIL
Higher education · bilingual programme scheduling
SBB CFF FFS
SIMBA MOBi · activity-based forecasting
unki
Travel tech · live itinerary engine
FOPH
Public health · pandemic policy
How we work
We work in small, focused engagements with the people who own the problem. No oversized teams. No black boxes. The model, the code, and the validation tools are yours.
Strip the problem to its essentials. What is being decided, what are the constraints, what would 'better' actually look like? Most of the value lives here.
We pair mathematical optimization, MILP, constraint programming, simulation, heuristics, with machine learning and LLMs where they earn their place: forecasting inputs, generating explanations, or letting domain experts edit the rules. The model fits the problem, not the other way around.
An optimizer is only useful if it stays trusted. We deliver software that planners can interrogate ('why this plan?'), that adapts when rules change without rebuilding from scratch, and that comes with the algorithmic tests to catch drift before it reaches a decision. You leave with a system that doesn't decay the day we leave.
Selected work
A selection of engagements across finance, energy, mobility, travel, and public health, each one a concrete decision made measurably better.
We built swissQuant a custom portfolio optimizer for risk-contribution-constrained allocation, a non-convex problem standard MIP solvers cannot solve to certified optimality. On real five-year MSCI World data, the heuristic alternative is up to 40% less risk-return efficient than our optimizer on 33-asset portfolios; in the most non-convex regime it drifts so far it returns a positive objective where the true optimum is negative.
Partner
swissQuant
Field · Year
Quantitative Finance · 2023
Our algorithm now plans the daily truck loads at Italmondo's Supernova Hub across Europe. It packs +7.7% more chargeable cargo than manual planning, saves roughly €1M per year, avoids about 1,000 t of CO₂, and removes the risk of illegal loads ever leaving the warehouse.
Partner
Italmondo · Supernova Hub
Field · Year
Logistics · Container Loading · 2024
Work together
You do not need a polished brief. Bring the planning, routing, scheduling, pricing, or allocation decision that is hard to make reliably. We turn it into a constrained optimization model, then add AI only where it makes the system easier to use, explain, or adapt.
We map the decision, constraints, data, and current workflow. The output is a clear modeling path, value estimate, and recommendation on whether to build.
We design the mathematical model, solve strategy, and planner-facing workflow. AI is used for explanations, forecasts, rule editing, or interface support when it earns its place.
Once the system is used, we recalibrate, compare recommendations with outcomes, update rules and formulations, and help your team take ownership.
Founders
Optimally was co-founded by Dr. Mikele Gajda and Dr. Tom Haering, two researchers who saw how modern AI could make operations research easier to explain, adapt, and maintain inside real organizations.
Co-founder
Optimization for logistics and transportation.
Co-founder
Large-scale choice-based optimization.
Get in touch
We'd like to hear about it. Short notes are welcome, tell us the problem and what 'better' would look like for you.
Pick a time that works for you
Write to uscontact@optimally.chWorking languages
English · Deutsch · Français
