Game & Level Designer
I design levels and mechanics that guide the player without a map: readable spaces, pacing that alternates tension and calm, and iteration until it plays right.
References, constraints, what the team can actually build in the time we have.
Top-down layout, points of interest, the mechanic that gives the level its loop.
Grey geometry in engine to feel dimensions, flow and dead zones before any art.
Watch players, cut what doesn't work, add verticality or calm where the pacing needs it.
Project 01
Step into the role of a detective and find the thief among 12 suspects. Each time you clear two of them you unlock new documents. Every case is unique, generated by AI.
A point-and-click investigation where you eliminate suspects, unlock evidence and name the thief — with a brand-new case every time.
Defined the concept, iterated three full game concepts, designed the five in-game documents, ran playtests and tuned readability.
Cut the interrogation phase and the pin board: one deduction loop, inspired by Guess Who?, that AI could actually generate reliably.
Scope to what the tech can deliver, and test reading load early — the documents went from 12 to 5, and lines from 12 to 8.
Develop the concept of a game entirely generated by AI, and bring it to life while keeping it engaging and enjoyable for the player.
AI can generate high-quality text and images. With that in mind we chose a visual novel format, and within it one genre stood out: the detective / deduction game.
We identified three families of detective games: deduction, investigation and contradiction-based. Deduction games were the easiest to pair with AI, so they became the base of our concept.
Notebook as the main toolPins and board connectionsText deduction with fill-in-the-blanks
The player interrogates two suspects, fills in their files by asking the right questions, then compares reports to spot inconsistencies. I spent most of my time on the interrogation phase and went through four versions of it.
.jpg)
.jpg)


We cut the interrogation and kept document analysis, with Papers, Please as the new reference. The player cross-checks 12 documents against suspects' statements on a pin board.
Red link = inconsistencyBlue link = correlation
Ten suspects, a notebook, and information that unlocks as you clear suspects. We cut the documents from 12 to 5 — the player analyzes them to incriminate some suspects and clear others.
I listed the documents that help solve a real theft case (ChatGPT first, then verified with my own research), studied real-life examples, and selected four that leave room for humor and are feasible with AI.
Criminal recordID cardInternet historyBank statement
Before the second version of the documents I built playtest sheets for each one, then asked several players a series of questions to find what to improve.
The game is mostly reading, so I tested fonts: a table where each line used a different typeface, and players told me which was easiest and nicest to read. The chosen font got a slight transparency to mimic dried ink.
Cleaner layouts, less noise, the dried-ink font — ready to be filled by the AI for each new case.
Project 02
You play a South American archer on a quest to lift the curse of an Aztec temple. Explore its depths and face formidable enemies to complete your mission.
One temple hall, four crystals, rising enemy waves — and a balcony that turns a flat arena into a vertical playground.
Designed the main level end to end: plan, gameplay loop, blockout in engine, two rounds of iteration, final layout.
Add a balcony carrying two of the four crystals. It fixed the flatness and pushed players to chain jump and dash.
A blockout shows dead zones no plan can: the first version had whole areas players had no reason to visit.
I was responsible for the levels and the gameplay mechanics. Most of my time went into the main level, so that's the one I break down here. On the tutorial level I mostly shaped how it unfolds rather than its layout.
Building on the team's concept art, and given the one-month deadline and our priority on fun, we settled on the interior of a temple: one large square room where the whole level happens. From there I conceptualized the level on my own.
Taking the lore and the 3Cs into account, I proposed a first layout with a simple loop that programmers could build and artists could dress cheaply: destroy crystals to deactivate the stele and end the level. Each crystal triggers an enemy wave; you can only break the next crystal once the wave is cleared, and difficulty rises with every crystal.
SteleEnergy crystalsEnemy spawnsPillarsFountainPlayer arrival
Once the plan was approved I built the first version in engine. The blockout let me feel dimensions and the space needed for smooth, dynamic movement — and it revealed a problem: large areas of the room gave the player no reason to go there.
Playing from the player's perspective, I spread the four crystals into the four corners to expand the playable area. It worked — but after testing, the level still felt flat. The answer was verticality, which would also exploit the character's mechanics.
A balcony runs around the hall with two of the four crystals on it. The player has to climb to break them, which kills the flatness — and encourages combining jump and dash to leap off the balcony and escape enemies.
Project 03
Take a seat as a galactic food delivery driver, working for a company that tends to misplace orders during intergalactic travel.
A die-and-retry across factories, minefields and falling debris — every trap, checkpoint and menu coded in Unreal Blueprints.
Programmed the whole game (traps, checkpoints, health, saves, keys, menus, collectibles) and assembled and tuned the level.
Rebuild the blockout around a tension curve: tutorial → factories → calm → minefield → calm → final gauntlet.
In a precision game, foreground clutter and off proportions kill the feel — and empty space needs a reason to exist.
I programmed the entire game in Unreal Engine with visual scripting, assembled the level design pieces, and tuned them so the game was both challenging and fun.
Having used Blueprints before on a smaller scale, I volunteered for programming. To scope the work, we first listed every level design element the player would interact with — and that I'd need to code.
I built each element with the help of tutorials and forums, deepening my Blueprint skills along the way. We tuned the parameters on a basic blocking setup — not the final level, but enough to feel the mechanics.
Then we tried the features on the first blockout of the real level to judge proportions, placement, length and more.



V2 guides the player subtly and balances challenge with rest. Here is the difficulty curve across the level:





Some sections changed again after feedback. I added a radioactive zone and falling objects to diversify the challenges. Beyond level elements I also programmed checkpoints, health, the save system, the key system for factory doors, menus and collectibles.
Project 04
An epic one-on-one battle. A wicked, jealous villain is about to attack a hotel full of grannies — but the magnificent Super-Épade has arrived to protect them.
A playable two-player brawl in three days: one hero, one villain, a hotel full of grannies to defend.

Brainstormed the gameplay mechanics and produced the hero's complete animation set.
My first hands-on game design work — the project that pulled me from 3D and animation toward design.
In a jam, ship what's playable: not every idea fit in three days, but the animations did.
In this group game jam I brainstormed the gameplay mechanics and created all the hero's animations. It was my third year of bachelor's, when we were mostly focused on 3D and animation — and this is where my interest in game design first sparked. Some ideas couldn't be developed in three days, but the animations shipped.









I'm Yanis Khier, a game and level designer trained at Ynov Campus Bordeaux. Video games marked my childhood — endless rounds of Bloody Roar 3 against my older brother, then Call of Duty's zombie modes, Titanfall's pilot-and-titan rhythm, Assassin's Creed's free-running cities, Arma 3's slow-burn realism.
What stayed with me wasn't just the fun. It was noticing why a level flows, why a fight feels fair, why one small detail makes a world believable. That curiosity became my job: I now build spaces that guide players without a map, and I iterate until they play right.
I'm a finisher. I see projects through, I'm exacting about the player's experience, and I've learned in kitchens and on teams that listening and clear communication get better results than being right alone.
Master's project: design and development of an AI-generated visual novel, Guilty or Not.
Master's project: a dynamic TPS in an Aztec universe, Darkest Canopy.