sleektiki.ai

Matthew Board

Technical Artist

02 / Projects

Selected Work

A selection of games, art, installations, and AI tools.

MAUDE

Multi-Agent Unified Dispatch Engine — a local AI assistant running entirely on the NVIDIA DGX Spark. Coordinates multiple specialized models: Nemotron for reasoning, Codestral for code generation, LLaVA for vision, and Mistral for general tasks. Features voice interaction with Whisper transcription, RAG-powered memory via vector database, autonomous sandboxed code execution, Telegram bot integration, and a client/server architecture over Tailscale VPN. Fully private — all inference on-device.

NemotronCodestralLLaVAWhisperTailscaleSQLite

MAUDE Mobile

A native iOS and Android companion app that connects to MAUDE on the DGX Spark over Tailscale. Seven modules in one app: AI chat with tool execution, full-duplex voice calls with live waveform visualization, SSH terminal via xterm.js over WebSocket PTY, web browser proxied through the Spark, Telegram message viewer, shared file manager with camera upload, and settings with theme switching. Built with React, Ionic Capacitor, and Tailwind CSS — a single TypeScript codebase compiled to native iOS and Android apps.

ReactCapacitorTailwindWebSocketOpus CodecPersonaPlex

Tessera

An AI-powered academic literature review platform that searches five major databases — Semantic Scholar, arXiv, OpenAlex, CrossRef, and PubMed. Generates structured paper summaries and cross-paper literature syntheses, visualizes citation networks with relationship classification, and extracts findings, methods, gaps, and conclusions into a searchable knowledge base. Export as BibTeX, JSON, or a deployable static website.

Next.jsReactSQLiteLLMSemantic ScholararXiv

Article-Gen

A multi-modal AI content generation platform that creates professional articles with AI-generated images and publishes directly to Substack. Auto-generates articles from trending news topics, creates three contextual images per article via Flux.1-dev, adds AI image captioning with LLaVA vision model, and supports custom prompt mode for full creative control. Writing style is fine-tunable via LoRA on Gemma 2 9B.

Gemma 2 9BFlux.1-devLLaVAStreamlitNVIDIA GPU

Pixelus

An AI-powered product placement platform that generates photorealistic brand scenes from a single product photo. Upload a product image, select a scene style, and Pixelus composites the product into a professionally lit environment using AI inpainting and depth-aware blending. Designed for e-commerce, social media, and marketing teams who need studio-quality product shots without the studio.

FLUXInpaintingDepth EstimationNext.jsRailway

Stillion AI LoRA Trainer

A FLUX LoRA training application running on the DGX Spark. Upload a dataset of images, configure training parameters like steps, learning rate, and rank, and fine-tune FLUX image generation models through an intuitive Gradio interface. Trained models can be exported and used for custom image generation workflows. Built for the Stillion & Board collaboration.

FLUXLoRADGX SparkGradioPython

Arch-Viz

AI-generated photorealistic architectural visualizations for real estate marketing, hospitality branding, and large-scale development presentations. Beachfront residences, urban towers, hotel interiors, and desert resorts, all from text prompts.

Flux 2 DevArchitectural VisualizationPrompt Engineering

Machina Prima

An exhibit concept series generated using a custom marker-mech LoRA trained on 15 hand-drawn mechanical sketches. The model learned the marker rendering style and was prompted to envision spaces for a futuristic art exhibition — from gallery entrances and sculpture halls to immersive projection rooms. Each of the 11 rooms was then converted into a pro-quality 3D Gaussian splat via the World Labs API, auto-cleaned to remove floaters and artifacts, and rendered as navigable 360° panoramas.

FLUX LoRAGaussian SplattingWorld LabsConcept Art3D

Input Streamliner

An Unreal Engine 5 editor plugin that generates complete multiplatform input configurations from natural language descriptions using a local Ollama LLM. Describe your controls conversationally and get a full input system — keyboard, mouse, gamepad, and mobile touch — with one click. Runs entirely local, no API keys or cloud dependency.

UE5C++OllamaEnhanced InputSlate UI

DataTable Streamliner

An Unreal Engine 5 editor plugin that generates fully populated DataTable assets from natural language descriptions using a local Ollama LLM. Describe your data — "12 fantasy weapons with name, damage, weight, and rarity" — and get a complete DataTable with an AI-defined or existing struct schema. Preview rows before generating, append to existing tables.

UE5C++OllamaDataTablesSlate UI

Kamodo Steve: Janitor on Fire!

A solo-developed indie game published on Steam, built in Unreal Engine. A janitor discovers the source of the poison hurting his town at a local chemical plant. Inspired by the same true story as the film Dark Waters. Features co-op multiplayer, custom character animation, hand-modeled environments with foliage systems, and a full boss battle sequence. All art, code, and design by a single developer.

Unreal EngineMayaSubstanceCo-opSteamSolo Dev

Stories from the Cores

An educational game where players analyze evidence from beneath the ocean floor to unlock dramatic stories. Join science bot Rovee on expeditions with the International Ocean Discovery Program — drill core samples, compare layers, investigate hypotheses, and piece together the stories hidden in the Earth’s geological record. Published on Steam for Windows, Mac, and Android. Also deployed as a museum kiosk interactive installation.

Unreal EngineZBrushMayaSubstanceSteamMuseum

Necessary Illusions

First collaboration between Stillion & Board after six years working together, exhibited at SPRING/BREAK Art Show at Skylight Culver City, Los Angeles in February 2022. Stillion’s paintings of face jugs, poppies, and houseflies evoke decomposition and the animation of inanimate objects — referencing African American pottery from 19th-century South Carolina. Board’s custom software manipulates and animates the imagery through code and algorithm. Featured in the “HEARSAY:HERESY” themed exhibition alongside 50+ curated shows.

PaintingCustom SoftwareInteractiveInstallationLA

Astronaut

A character study exploring realistic skin and hair rendering in Unreal Engine 5. Head textures were built layer-by-layer in Substance Painter — from blood through skin layers to a final makeup pass. Hair was crafted in XGen with guides. Rendering leveraged Metahuman eye shaders and subsurface profiles, with a ZBrush cavity map breaking up surface roughness and doubled-up normals for extra skin detail. The suit was retopologized in Modo, sculpted in ZBrush, and painted in Substance Painter.

ZBrushSubstance PainterUE5XGenModo

Iggy Pop

A digital likeness inspired by seeing Iggy Pop perform live in Chicago in 2010. Began as a personal character study and evolved into a real-time rigged model. In 2016, AST Studios licensed the character model and textures for the Oneohtrix Point Never & Iggy Pop music video “The Pure and the Damned” from the motion picture Good Time, which debuted at Sundance 2017. Recently reworked for UE5 with subsurface shading, strand-based hair in XGen, repainted textures in Substance Painter, and path-traced rendering.

ZBrushXGenKeyshotSubstanceUE5Sundance

04 / Contact

Get in Touch

Have a project in mind, want to collaborate, or just want to say hello?

Email: matt at sleektiki dot ai

About

About Me

Indie game developer, technical artist, and creative AI researcher exploring the intersection of artificial intelligence and 3D content creation. From shipping games on Steam to building local AI systems on NVIDIA hardware, my work bridges traditional modeling and animation pipelines with emerging AI-driven workflows.

These cross-disciplinary experiences inform everything I build — preparing the next generation of digital creators to work alongside AI tools rather than be displaced by them.

Disciplines

Creative AIGame DevTech Art3D ModelingAnimation