The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
I Fine-Tuned a Model. The Failures Taught Me More Than the Winning Score. What a classification win over GPT-4.1—and failures on unfamiliar regulatory questions—taught me about building useful domain intelligence.
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Challenges
View all
Related posts
Engaging website for a New York–based architecture and construction studio, balancing bold typography, immersive imagery and a minimal editorial layout.
What's your one-rule brand? Tell me.
Most jewelry brands photograph the object. This one photographs the grip.
AQUILIFER — a brand I built around one idea: jewelry as something carried, not worn.
Case study + full site now live.
The name comes from Roman military rank. The aquilifer carried the standard — the one thing the legion couldn't afford to drop.
That single word killed my first direction before I'd finished it.
No polish. No faceted stones.
Hand-cast, asymmetric pieces, tool marks left visible.
Photography went extreme macro, unretouched skin, hands gripping the face instead of posing near it.
Color system: two darks instead of one dark + accent.
#1E1E1E and #27331A let the brand shift register — campaign world to product world — without ever going light.
#D9DED4 stays structural only. Never a fill.
Motion rule on the site: diagonal masks only. No rotation, anywhere.
A spinning object reads as being shown off. This brand is about weight, not display.
Same angle, same easing, top to bottom.
One idea, held all the way through — name, object, photography, palette, motion.
That's the whole pitch for how I work.
Absolutely love your works!
🧠 GROWTH TIP 06/30: DON’T START WITH WHERE. START WITH THE FEELING.
Most travel apps start with one question: “Where do you want to go?”
That works when the user already knows the answer. But a lot of travel discovery starts earlier, with a much less specific thought: I want to go somewhere, I just don’t know where yet.
That idea became Traveling Up, a travel randomizer concept.
Instead of asking users to work through dozens of filters, the product starts with just three signals:
🌍 how much time they have
🎭 how they want the trip to feel
☀️ what kind of climate they prefer
The destination stays a surprise.
Hypothesis: removing destination selection from the first step could move users faster from vague travel intent to genuine interest in one specific trip.
What I would test:
• Destination first search vs Mood first randomizer
• Randomize start rate
• Destination save rate
• Plan this trip CTR
• Time to first interesting destination
💡 Growth insight: in discovery products, uncertainty is not always friction. Sometimes it is exactly what creates curiosity and moves users toward the next action.
This concept came to me after a particularly inspiring trip. And I think it is the first mobile concept I have created purely for myself, rather than for a client project, in roughly three years.
So this one has a little more freedom, art direction, and product experimentation in it.
06/30 complete.
If you enjoy seeing mobile product experiments like this alongside the growth thinking behind them, follow along. And if this direction resonates, a reaction helps me understand which ideas are worth exploring deeper next.
#MobileAppDesign #TravelApp #TravelAppDesign #MobileUX #UXDesign #ProductDesign #GrowthDesign #TravelTech #InteractionDesign #UIDesign #ProductStrategy #MobileProduct #AsolDesign
Perfect
