I turn creative ideas and practical problems into working software, instruments, and interactive experiences.
My background combines 15+ years in creative strategy, advertising, digital production, and product storytelling with an engineering education and hands-on product development. I work across web applications, local AI, Windows utilities, music technology, and games, taking projects from concept and interface design through implementation, iteration, and validation.
Based in Metro Vancouver, Canada.
Portfolio · LinkedIn · YouTube · Email
Explore AI image styling through a physical gesture. Hold two L-shaped hands in front of a webcam to reveal an AI-styled version of the scene inside a movable window.
How I built it: I developed the concept through iterative prototyping with Codex and hands-on webcam testing. Python, OpenCV, and MediaPipe handle capture, hand tracking, and compositing; local ComfyUI workflows style the full camera image with FLUX.2 Klein 4B or Qwen Image 2.1 Turbo. Styling the whole scene before applying the hand-controlled mask keeps the model's scene context steady as the window moves. A separate desktop UI process stays responsive while inference runs.
Status: Windows prototype with local inference. AI image updates are asynchronous and slower than camera tracking.
Demo and gallery · Development story
Make typing practice feel like playing. Words slice flying food; sentences prepare ingredients and serve restaurant orders. Adjustable pace, finger guides, stories, and browser-local records support short practice sessions.
How I built it: I directed the game concept, mechanics, visual design, and iteration with Codex assisting implementation and debugging. TypeScript, Three.js, and Vite power the game. Blender, Blender MCP, and Python produced 50 editable 3D foods with whole and cut models, plus the kitchen. GPT image generation supplied reference art and illustrated cards; Web Audio synthesizes the music and effects. Development included automated checks, browser playthroughs, and multi-view asset reviews.
Status: Published browser game. No account required; gameplay makes no AI requests.
Play TypeSlasher · Explore the 3D foods · How it was made
Bring musical performance into interactive scenes. This early Unreal Engine plugin explores connecting MIDI controllers, and eventually audio, to lights, motion, materials, and gameplay.
How it is being built: A C++ runtime plugin handles MIDI discovery, connections, normalized events, and Blueprint access. A native editor monitor, synthetic test inputs, and a small light-and-motion rig make individual behaviors inspectable. Initial testing uses a Maschine MK3; pad input and synthetic cube motion have been observed.
Status: Early implementation. Physical knob validation, audio processing, general mappings, and a playable packaged demo remain pending.
See AI coding limits without breaking your flow. A Windows tray companion and compact fish-themed widget show Codex and Claude Code quota, reset times, and whether readings are fresh.
How I built it: I shaped the product, interface, and desktop behavior with Codex assisting implementation, review, and validation. C#, .NET 10, WinUI 3, and the Windows App SDK provide the native interface. Separate adapters read the local Codex App Server and Claude Code status-line observations, then normalize them for both views. Provider authentication stays with the existing tools.
Status: Windows x64 portable beta candidate; a public release has not yet been published.
Architecture · Development story
Use a photograph as a starting point for sound design. LUMEN turns image data into playable wavetables, with a simple performance view and a deeper synthesis editor.
How I built it: I defined the instrument concept, interface, interaction behavior, and validation approach, directing Claude Code through development. C++20, JUCE 8, and CMake support the synthesizer, image-to-tone engine, modulation, MIDI controls, effects, and 32 factory presets. Validation includes automated audio measurements and plugin checks.
Status: Released Windows standalone application and VST3 plugin. Image-to-tone processing is deterministic and local, with no generative AI model at runtime.
Turn visual inspiration into editable musical phrases. Lumena converts an image into MIDI that can be developed further in a music production workflow.
How I built it: I developed the image-to-music concept with Claude Code and Codex supporting implementation and iteration. The standalone C++17 library combines image brightness and color analysis, musical key selection, a weighted Markov melody engine, phrase shaping, and a MIDI writer. Fixed images, settings, and seeds produce reproducible output.
Status: Implemented and tested standalone library with an end-to-end command-line demo. Algorithmic music generation runs locally without an LLM.
Move from product information to a coherent marketing campaign. Markitekt turns product details and imagery into strategy, creative direction, social calendars, captions, campaign visuals, and video storyboards without requiring users to write prompts.
How I built it: I independently conceived the product and designed its experience and creative workflows, using Claude Code and Codex for development support. Claude API supplies reasoning and OpenAI supplies image generation. Next.js, TypeScript, Supabase, Stripe, Resend, and Vercel connect the interface, data, payments, email, and deployment.
Status: Live beta with a private source repository.
Learn Git by seeing what commands do. An isometric browser adventure turns repositories, staging, commits, branches, and merges into guided quests and visible state changes.
How I built it: I designed the learning structure, game interactions, and art direction. JavaScript and Three.js support the world and simulated Git workflow; Claude Code and Codex assisted implementation, debugging, and interface refinement.
Explore an AI assistant that works with local knowledge. This local-first project brings together conversation, voice workflows, and retrieval over locally stored information.
How it is built: A React interface connects to a Python/FastAPI backend. Local model workflows and SQLite FTS5 support offline retrieval, including an implemented Wikipedia import and indexing pipeline.
Status: Development project. The repository now uses the name SCRAP; optional internet search is described as future work.
- MeowSushi: A browser cooking game serving sushi to feline customers. I created the concept, mechanics, visual and audio direction, using Claude for planning, Claude Code for development, GPT Image for visual assets, Suno for city soundtracks developed from my demos, and ElevenLabs for sound effects.
- Alpha Jets: A browser hero-shooter experiment where letters and numbers become characters. I developed the concept, character design, game systems, and balancing with Claude, vanilla JavaScript, and Canvas.
- AI workflow tutorials: Practical demonstrations of generative media, model selection, and reusable ComfyUI workflows, including local image-to-video production.
I start with the user problem, define the behavior and constraints, and build a small working version. I use Claude Code and Codex where they fit the project, then inspect the implementation, test behavior, review the creative output, and refine the result.
My contribution spans product concept, requirements, UX, creative direction, implementation decisions, debugging, and validation. I document the development process so the choices behind the finished work are visible.
AI plays different roles across these projects: runtime inference in GesturePortal and Markitekt, development assistance in tools such as LUMEN and TokenFish, and asset creation alongside conventional 3D modeling in TypeSlasher.
| Area | Tools and techniques |
|---|---|
| Web and interactive | TypeScript, JavaScript, React, Next.js, Three.js, Canvas, Vite |
| Desktop and audio | C#, .NET, WinUI 3, C++, JUCE, CMake, DSP, MIDI |
| AI and data | Python, FastAPI, ComfyUI, local LLMs, RAG, structured outputs, PostgreSQL, Supabase, SQLite FTS5 |
| Creative production | Blender, Unreal Engine, Adobe Creative Cloud, Figma, Suno, ElevenLabs |
| Development and validation | Codex, Claude Code, Git, GitHub Actions, automated tests, browser playthroughs, visual and audio review |
- Founder and AI Creative Technologist at Kazi Ahmed Media, combining product building with creative production.
- Former Content Strategist at Niricson (September 2022 to October 2026), translating AI and robotics products into positioning, demonstrations, technical education, and customer-facing content.
- Creative and marketing experience at Leo Burnett and Grameenphone / Telenor, with confirmed multidisciplinary team leadership across agency, brand, web, and digital production roles.
- Suno creator-program consulting internship and a separate contractual Champion Ambassador term (July 2025 to July 2026). Managed 45+ creators and their program alongside the program director.
- Human+AI music creator with 25K+ Suno followers and 10M+ Suno Hooks views, as of October 2026.
- MIT Professional Education: Applied Generative AI for Digital Transformation.
- BSc: Electronics and Telecommunications Engineering, North South University.
- Post-baccalaureate diploma: Digital Marketing, Thompson Rivers University.
Portfolio · LinkedIn · Suno · YouTube · Instagram · TikTok · Reddit
Contact: pixelsncodes@gmail.com
Project descriptions and status reviewed October 9, 2026.