Mentoring 250+ across 11 faculties
The single moment that anchors this post is a biology student in 2024 — second year, no design background, three hours into a Saturday Figma workshop — turning her laptop around to show me a small interface for her research lab's specimen catalog. It was simple, well-typed, used a real grid, and was honestly better than what most of her CS-major peers would have produced at the same point. She'd never opened Figma before that morning.
That happened often enough across three years and eleven faculties that I stopped being surprised. The patterns that actually predict whether a room learns the material aren't the patterns I expected when I started.
This is what holds. And what doesn't.
the rooms
A loose taxonomy of the eleven faculties I've worked with. The diversity matters because it's the data the patterns came from.
- STEM majors (Industrial Engineering Catalyst, Chemical Engineering KMTK) — analytical, comfortable with abstraction, often surprisingly good at the technical-fluency aspects of design but cautious about subjective choices
- Pure sciences (BEM Biologi, Geography Study Club, Gama Cendekia) — research-trained, asking the right questions but slower on the visual-decision-making side
- Humanities-adjacent (English Competition SV, Climate Change Initiative) — strong communication instincts, faster on the what is this for question, less confident on tool mechanics
- Religious / community (Keluarga Muslim Teknik, Ramadhan Di Kampus LDK) — high group cohesion, design work that has to land for shared community use, different stakes
- ICT student community (Komatik UGM, GEMASTIK mentor sharing) — closer to home turf, faster on tools, less benefit-of-the-doubt on framework claims
The category that mattered least was technical background. The category that mattered most was whether the workshop output had a real downstream use. A bio student designing for her actual lab outpaced a CS student designing a fictional resume because the bio student had a real user in mind.
what holds across rooms
Three patterns that worked in every faculty I taught.
One decision per slide. The structural rule the whole curriculum runs on. A slide that asks "what do you want this to communicate?" works in any room. A slide that lists eight Figma features at once works in zero rooms. Compression of cognitive load forward beats coverage backward, every time.
Show, don't explain. A live walkthrough of building one component in real time, with the audience following, beats any number of theoretical slides about hierarchy or contrast. The closest engineering analogue is show the code being written, not show the finished code. The verb matters: building is what teaches, finished is what intimidates.
Make a real thing in the session, not a fake exercise. This is the most under-rated principle. A workshop that ends with the student holding a thing they actually need is worth ten workshops that end with a Figma file labeled "exercise.fig." The bio student's specimen catalog wasn't an exercise. It was a real lab need someone was finally building.
what doesn't transfer
Three patterns that worked in some rooms and failed badly in others.
Design-history references. Mentioning Bauhaus, Swiss design, or any of the canonical references lands well in design-trained or humanities rooms. It lands flat or actively counter-productively in STEM rooms — students hear "Bauhaus" as a fact they're now expected to memorize rather than as a frame for the principle. I cut almost all canonical-design references for STEM workshops; I keep them for humanities-adjacent ones.
Encouraging-language style. "Make it your own, trust your instincts" works in rooms with high creative confidence (English Competition, BEM Biologi) and falls flat in rooms where students need more structure, not less (Industrial Engineering, Chemical Engineering). For the structure-needing rooms I lean on prescriptive frameworks — use this 12-column grid, use this 1.25x scale ratio, use this color palette unless you have a specific reason not to. The "make it your own" framing comes later.
The clickbait Figma examples. I tried using highly-stylized, design-Twitter-style Figma examples in early workshops. They impressed but didn't teach. The students would copy the surface aesthetic without understanding the underlying decisions. I've moved to boring-but-correct examples — visibly defensible choices over visibly trendy ones.
the second-order effects
The compounding I didn't expect:
Left the workshop and rebuilt my org's deck the next day.
Students who showed up to my workshops came back as mentors for the next cohort. The Industrial Engineering Catalyst session in 2024 produced two people who led their own design workshops in 2025. The Komatik 2026 GEMASTIK session was led partly by a student who'd attended one of my 2024 sessions.
Decks I'd built for one session got adopted by orgs for their own training. Design files I produced as workshop examples ended up as templates other people used for actual projects. The artifacts of teaching turned out to have a longer half-life than the teaching sessions themselves.
The two-step compounding pattern: teach 25 people directly; some fraction of those 25 teach 25 more each, using the same materials, sometimes the same slides. The 250+ direct number understates the actual reach by some unknown multiplier.
what i'd do less of, more of
Less of: long sessions. The 3-hour format works, but I've tried 4 and 5 hour sessions and the attention curve drops too steeply after hour 3. The marginal hour adds frustration, not learning. I now decline sessions longer than 3 hours unless they're structured as two separate 3-hour blocks with a real break between.
More of: small group sessions (8-15 people) instead of large lecture-style sessions (40+). The retention difference is dramatic. Small-group sessions where every student gets at least one minute of direct feedback retain 3-4x more learning than large sessions where feedback is generalized. The cost is throughput — I can teach 250 people in lectures or 80 people in small groups in the same time — but the per-student outcome is much better with the small format.
More of: follow-up. The single biggest improvement to my workshops over three years was adding a 2-week follow-up — a check-in on the actual thing the student was building, not a check-in on the abstract skill. The follow-up converts more participants from attended the workshop to actually using the skill.
why i keep saying yes
Three years in, the invitations don't stop. I've turned some down. I keep saying yes to most of them. The reasons:
- The mentorship work itself is the most efficient teaching I can do for my own engineering practice. Explaining a concept to a non-engineer 50 times strips the concept down to its essential parts. My internal engineering documentation is 4x clearer than it was three years ago, entirely because of these sessions.
- The compounding is real and visible. The students who came back as mentors are evidence that the work doesn't decay when you stop showing up.
- The unit economics, if I'm being honest, are good. Three hours of teaching produces hundreds of hours of downstream impact. Almost no other professional activity I do has that ratio.
The work isn't charity. It's investment in a long-term audience and a long-term reputation, with the side effect of being one of the most satisfying things I do professionally. The "why I keep saying yes" answer compresses to: the compounding works both directions — the students get the workshop, I get the discipline of explaining clearly, the room gets the kind of cross-faculty bridge that wouldn't exist otherwise.
This is the most durable single body of work I've built in three years. Not the enterprise platform, not the freelance projects, not even the public engineering work. The 250 conversations across 11 faculties.