Competencies:
DULDAM
Money isn't lost when food is thrown away. It's lost when too much is made.
With a team of six, I rebuilt our food-waste proposal for institutional caterers around a single meal's economics, then built the website and a desktop simulator that let a caterer see its own numbers before agreeing to anything.
A caterer earning about KRW 180 on a KRW 5,000 meal will not change its serving line for an environmental slogan. It will for a measured cost, tested free at one site.
- KRW 129
- Recoverable per meal, against about KRW 180 operating profit (simulated, 800-meal site)
- 97%
- Of the saving comes from making less, not from smaller leftovers
- KRW 0
- Pilot cost to the caterer: no equipment, no extra staff
- 4 weeks
- One site, from baseline to a before-and-after report
Windows 10/11 (64-bit) · ZIP, 158 MB · v0.4.1
- Pilot Offer
- Team Project
A thin margin with a predictable leak
Contract catering runs on margins in the 3% range. In the first half of 2026 the three listed Korean caterers reported operating margins between 2.0% and 6.9%, which leaves less than KRW 200 on a KRW 5,000 lunch. That margin leaks from the same three places every day: food cooked but never served, food left on the serving line, and food left on the tray.
Most waste campaigns start at the tray-return counter. By then the food has already been bought, washed, cooked and served, so the money is gone. We measure the three points separately, because each one recovers a different kind of cost.
The number we took out
Cutting plate waste only saves the disposal fee: about KRW 70,000 a month at an 800-meal site. Making 40 fewer meals a day saves about KRW 2.2 million in ingredients. That is why 97% of the effect sits in production, and why the pilot puts its effort into measurement that a dietitian will trust enough to change the plan.
There was a tempting shortcut. Convert the drop in plate waste into portions and multiply by ingredient cost, and you get an impressive KRW 1.86 million a month. It is food the caterer has already paid for, so it never comes back. We left it out of every total, and the simulator flags it as a common error.
Two changes that cost nothing
Neither change needs new equipment. Rice moves later in the serving line, so diners see the whole meal before deciding how much staple food to take. And one sign changes: instead of "Please don't leave food", it says "You can always come back for more", which makes a smaller first serving feel safe.
- Week 1BaselineWeigh production, serving and returns as things run today
- Week 2Serving redesignNew serving order and the refill sign
- Week 3ProductionDay-of-week forecasts fed into the production plan
- Week 4VerifyCompare with the baseline against agreed criteria
The measurements then feed the change that matters most: day-of-week production, so that a quiet Friday stops being cooked for like a busy Tuesday.
Making yes easy
An earlier version of the offer charged a small fee per site. We dropped it. The pilot is now free, the caterer gives one site and about 26 minutes of staff time a day, and the success criteria are agreed before it starts: plate waste per person down at least 10%, lower production surplus, and no rise in stock-outs or complaints.
To make the first meeting concrete, I built the DULDAM website, with an AI diagnosis that turns one sentence about a cafeteria into simulator inputs, and a desktop simulator for Windows and Mac that splits disposal savings from ingredient savings.
- 1Side dishes
- 2Main
- 3RiceNow third
- 4Soup
You can always come back for more
Source: DULDAM proposal, p.5, serving-order redesign and refill message.
Where it stands
- Researched food waste in institutional catering in the ESG Management Team of the Korea University Law & Policy Society
- Rebuilt the proposal around single-meal unit economics, with a free four-week pilot
- Launched the DULDAM website: AI diagnosis, simulator and operations dashboard
- Released the desktop simulator for Windows and Mac (v0.4.1)
- Sending the proposal to 15 institutional caterers, split across the six team members
- Run the first four-week pilot at one site
- Final operator-facing business proposal by 26 December 2026
KRW 69,696 / month vs KRW 2,200,000 / month. Simulated for an 800-meal office cafeteria, 22 operating days.
Benjamin Franklin said a penny saved is a penny earned. At a 3.6% margin it is worth far more: every won a caterer stops wasting is worth about 28 won of new sales.
My role
I co-developed the business model with my team, rebuilt the cost logic around a single meal, and built the website and the desktop simulator myself with AI coding tools.
Skills used
Unit economics, Commercial judgment, Pilot design, B2B proposition, AI-assisted prototyping