Game designer · Props artist · London

I design game systems and build the props that sell them.

Game designer and props artist. I led the weapon design team at Akiyam Solutions, and I prototype in Unity and Unreal Engine 5 and model hard-surface props in Blender.

See case studies ArtStation

Case studies

Professional and personal projects in depth: how they were designed, what I owned, and what I learned.

Venom skin set on five weapons: SMG, shotgun, pistol, assault rifle and marksman rifle, with teal acid camouflage

Weapon design · Team lead · Released on Steam

Weapon systems at Akiyam

I led an 11-person weapon team on ASSASSIN: The First List, split between artists who modelled weapons and developers who made them work in game. I set the modelling standards (polygon budgets, rig setup, part naming), ran the review step between the two groups, and moved the team's files off Discord to secure cloud storage.

I also pitched a themed DLC weapon pack and balanced it for single-player: higher damage, but scarce ammo and close range, so it works as a panic button rather than replacing the base arsenal. The Venom skin set shown here was designed for that pack and never shipped.

Role: Weapon Design Team Lead
Studio: Akiyam Solutions
Team: 11 people

Read the case study →

Low-poly 3D printer model in Blender beside its UV layout

Systems design · UE5 · Prototype, paused

3D Printing Simulator

Run a 3D printing business from your desk. Take orders by email, print and deliver them, then spend what you earn on better printers and materials. Start alone, grow into a team, and decide whether to stay legal.

I designed and programmed the core order-to-delivery systems and modelled the printers.

My role: Core systems, 3D models
Engine: Unreal Engine 5
Team: 2 people

Read the case study →

University projects

From my Postgraduate Diploma at Brunel: a four-stage prototype project, two solo games and concept design. Use the arrows to browse, and click any clip or screenshot to open the full video and gallery.

University · Four prototype stages · Team, then solo

Escape from Candyland

The brief was one word: Wonderful. Our answer was a world made of sweets, full of hidden paths and tricks that make players stop and ask how to get through.

The first two stages were team work. We built a paper prototype, then a 2D version in Clickteam Fusion, where I designed and built two of the five levels: a gentle opener that teaches movement, and a harder level of traps and branching routes.

The last two stages were solo. I took my level into Unity as a 3D vertical slice, then rebuilt it from scratch for the high-fidelity version, adding loot boxes that hide weapons, land mines for dealing with enemies tactically, and key items that unlock doors and mechanisms.

Stages 1–2: Team of 5
Stages 3–4: Solo
Tools: Paper, Clickteam Fusion, Unity

Personal projects

Unreal Engine work outside university: two systems from my rebuild of Skyward Pioneers, and an atmosphere study.

3D art

Props, materials, scenes and branding, modelled and rendered in Blender. Open a project to see clay passes, wireframes and layout plans.

About

I'm a game designer and 3D props artist in London. At Akiyam Solutions I moved from 3D Modeler to Senior Game Designer to leading the weapon design team.

I hold a Postgraduate Diploma in Digital Games Theory & Design from Brunel University London, where I built prototypes in Unity both solo and in teams.

ExperienceWeapon Design Team Lead · Senior Game Designer · 3D Modeler — Akiyam Solutions

EducationPGDip Digital Games Theory & Design, Brunel University London

Tools

Unreal Engine 5UnityBlenderPhotoshop

LanguagesEnglish, Hindi, Gujarati, Sanskrit

← All case studies

Case study · Systems design

3D Printing Simulator

A first-person business sim where you run every step of a real 3D printing workshop: read customer emails, configure the slicer, load filament, watch the print, and deliver the finished part. Then hire a team to handle it while you grow the business.

My role: Core systems design & programming, 3D models
Engine: Unreal Engine 5, Blueprints only
Platform: PC (Steam)
Team: 2 people
Status: Prototype, paused

Design pillars

Four rules every feature had to pass. Inspired by Schedule I, PowerWash Simulator, Euro Truck Simulator 2, House Flipper and the Cura slicer.

Realism over complexity

Every step mirrors how real 3D printing works, without burying the player in settings.

Consequence over punishment

No game-over screen. A bad choice costs quality, money or reputation, and the player sees exactly why.

Depth through layering

Start doing everything by hand, then delegate one task at a time to staff and upgrades.

Architecture first

Systems are built so new features plug in without rewriting what already works.

Core loop

Every order feeds the next. Money buys better equipment, reputation brings more and better orders, and XP unlocks new materials, models and staff.

One order, step by step

Thirteen steps, about 5 to 7 minutes per order. Each one is a physical action in the workshop, and each one can later be handed to staff.

Take the job
- Read the email
- Accept it into the job queue
Slice
- Open the slicer app
- Pick the right model
- Configure settings
- Send to the printer
Print
- Load the right filament
- Monitor and fix failures
- Collect the part
Finish
- Remove supports
- Pack the box
- Hand it to the customer
Get paid
- Reward × quality × deadline

How an order is scored

Final reward = base reward × quality score × deadline multiplier. Mistakes reduce the payout instead of failing the order, so players learn from every job.

Quality	Score
Everything correct	1.0
Wrong colour	0.8
Wrong material	0.6
Wrong model	0.4
Supports off when needed	−0.2
Fast speed or draft layers	−0.1 each

Delivery	Multiplier
Before deadline	×1.1
On time	×1.0
Late (up to 50% over)	×0.8
Very late	×0.6
Not delivered	×0

Customer reaction	Reputation
90–100%: happy	+2
70–89%: neutral	0
50–69%: disappointed	−1
Below 50%: angry	−3 + complaint

Printer progression

Each level unlocks a better printer, and each printer opens new materials, which open new models to sell.

Printer	Price	Level	Materials
Basic FDM	£299	1	PLA
Mid FDM	£799	2	PLA, PETG, ABS
Advanced FDM	£1,499	3	All FDM plus TPU
Resin SLA	£2,499	3	Resin models only

Product catalogue

Beyond the 13 demo models, I specified a catalogue of 26 products, each with a model size, nozzle and bed temperature, filament and complexity. Filament ties the catalogue to the printers: new materials mean new things to sell.

Filament	First printer	Products	Examples
PLA	Basic FDM	17	Phone cases, miniatures, plant pots
PETG	Mid FDM	5	Camera mounts, prosthetic covers
ABS	Mid FDM	2	Replacement parts, containers
TPU	Advanced FDM	2	Wearables, kitchen gadgets

Balancing the economy

I modelled the economy in a spreadsheet before any playtest: order time from model size and complexity, material cost per filament, and profit per minute for every product on every printer, all inside the GDD's rule that orders pay £8 to £100. The first run showed one product winning everywhere. After one balancing pass, every printer has a healthy spread.

Printer	Best product before	Best ÷ median before	Best product after	Best ÷ median after
Basic FDM	Cookie cutters	5.6×	Miniatures	1.3×
Mid FDM	Cookie cutters	5.4×	Miniatures	1.2×
Advanced FDM	Cookie cutters	4.9×	Kitchen gadgets	1.2×

Print time scales

Every product originally took the same 20 minutes. Time now grows with size and complexity, so a typical order lands in the GDD's 5 to 7 minutes.

No single winner

A £250 cookie-cutter order broke the £100 cap and earned 5.6× the median. Repricing it and fixing prints nobody would bother with evened out every printer.

Upgrades that pay

Products that unlock on a better printer earn more per minute, so every upgrade is worth switching to.

Pacing

At the median rate, the Mid printer is affordable after about 1.3 hours of printing, the Advanced after about 3.7 and the Resin SLA after about 7.3.

Is crime worth it?

I ran the expected value of each shady stranger offer: what it pays on average once the chance of getting caught is counted.

Unlicensed job

3× reward with a 25% raid and a £500 fine. It only beats a normal job on orders over £100, and no order pays that much, so it never pays on average. Only gamblers take it.

Stolen filament

60% off with a 10% chance of a £300 fine. It only saves money on spools that cost over £50.

Black-market parts

Half price with a £1,000 fine if they fail. With a 10% failure chance they pay off on anything over £200, which makes a black-market printer tempting by design.

How I built it

Built entirely in Unreal Engine 5 Blueprints. Each order is a record that moves through a status pipeline, and the job queue drives all the work in the workshop.

Pending
Accepted
Slicing
Printing
Post-processing
Packing
Dispatched
Delivered / Failed

One function per task

The player and staff call exactly the same functions, so hiring a worker needs no new code. Staff simply do the player's job.

Self-contained printer

The printer actor has one entry point, StartPrint(). The slicer talks to it only through a single settings struct.

Swappable visuals

The print animation and the model preview are isolated functions, so the demo's simple versions can be replaced without touching the logic.

Data-driven orders

Orders are generated from data tables: 13 models, 20 customers, 9 email templates and three customer personalities, giving 3,240+ combinations.

Scoping the demo

To hit a playable demo, I split every system into a demo version and a full version, so each feature shipped small first instead of being cut.

In the demo

Full order generation
Slicer with 5 core settings
Printer with progress animation and status light
Simple filament stock check
Quality scoring
Packing and a simple customer reaction
Basic HUD and 3-panel email inbox

Full version

Temperature settings and physical filament spools
Print failures, bed levelling, first-layer checks
Support removal and saved slicer profiles
Filament and equipment shop
Staff, wages and boosters
Shady stranger events
3D model viewer and save system

Progress when paused

Phase 1, data layer: done
Phase 2, orders, inbox, day/night: done
Phase 3A, manual print loop: in progress
Phases 3B to 7: planned

Full-version systems

Designed in the GDD and planned for after the demo.

Staff

A print technician, a workshop assistant and a delivery driver, each paid weekly. Low morale means slower work and mistakes. Two missed wages and they quit.

Boosters

Coffee, energy drinks and pro boosters speed staff up, but the stronger ones cause a crash day if used too often. Unhappy staff refuse them.

Shady strangers

Once a week a stranger offers a deal: an unlicensed job for 3× reward with a 25% raid risk, cheap black-market parts that might catch fire, or stolen filament.

Print failures

Layer shifts, nozzle clogs and bed adhesion failures, plus a 30-second first-layer check. Premium filament and upgrades remove the risks one by one.

Printer model

Modelled in Blender for a game engine: a low-poly mesh with a packed UV layout, then a clean clay pass to check forms.

Low-poly printer mesh in Blender next to its packed UV layout — SM_Printermesh: low-poly mesh and UV layout

Clay render of the printer frame and print bed — Clay pass of the frame and bed

Credits

Co-developed with Nitish Kumar Panda, who designed the levels. Development is paused.

← All case studies

Case study · Weapon design & team lead

Weapon systems at Akiyam

I led the weapon team on ASSASSIN: The First List, a single-player shooter released on Steam. My job was to get a large roster of weapons designed, modelled and working in the game, and to keep 11 people moving in the same direction.

Role: Weapon Design Team Lead
Studio: Akiyam Solutions
Team: 11 people
Engine: Unreal Engine

The roster template

I designed the roster template the team worked from: weapons sorted into seven classes and four rarity tiers (Basic, Prime, Legendary and Extra), with magazine, fire rate and damage for each. The weapons we built in-house are under NDA, so this version is filled with 144 free-licensed weapons to show the same structure. Each class is built for a different situation, from close-range spray to one-shot precision.

Class	Weapons	Magazine	Fire rate (rpm)	Damage per shot	Role
Pistol	15	7–30	400–900	30–55	Reliable sidearm
Revolver	16	1–13	15–50	25–80	Slow, heavy single shots
SMG	33	13–64	600–1,100	20–45	Close-range spray
Assault	20	20–40	500–950	28–50	All-rounder
LMG	1	200	800	30–40	Sustained fire
Shotgun	28	2–8	Single–180	45–90	Burst damage up close
Sniper	31	4–10	Single–55	75–115	One-shot precision

Running the pipeline

The team was split between artists who modelled weapons and developers who made them work in game. I sat between the two.

Standards

Every weapon class had limits for vertex count and size, a required rig setup, and consistent part and bone naming, so any model worked with any developer's code.

A review gate

Artists delivered the model, textures and a spec file (name, type, poly count, notes). I checked each one before it reached a developer.

Secure file sharing

The team shared work over Discord. I moved it to cloud storage to keep unreleased assets from leaking.

Making the case with data

When downloaded models started replacing in-house work, I compared both side by side. Ours had smaller files and better performance, and cleaning up downloaded models took as long as building new ones. The studio went back to building in-house.

A DLC weapon pack

I pitched a themed weapon pack and balanced it for a single-player game. DLC weapons hit harder, but with scarce ammo and close range, so they work as a panic button for tough fights rather than replacing the base arsenal. The Venom skin set below was designed for that pack and never shipped.

What I learned

Replace a check before removing it

When I stepped back from reviewing every model, nothing replaced that check, and that's when problems slipped through. Now I hand over a process before I hand over the job.

Write everything down

The tracking sheets I kept are what convinced leadership. I now document references, decisions and sources as I work, not afterwards.