davidzha712/ComfyUI-Fizgig-H3-Still
True single-frame stills from MiniMax H3 in ComfyUI: a one-frame latent and a clean decode, at any size. No extra models.
Senior ML Engineer & Data Scientist | NLP & CV | Published Researcher
Repositories
davidzha712/ha-jev-autopilot
Home Assistant integration: TypeSafe Jev decides what each room needs; deterministic code acts, with phone confirmation for risky devices.
davidzha712/narrated-web-app-demo
Claude Code / Codex / Gemini skill: narrated voice-over demo videos of web apps (incl. SSO-authenticated), built on splitbrain/ndemo
davidzha712/FlashVSR
[CVPR 2026] Towards Real-Time Diffusion-Based Streaming Video Super-Resolution — An efficient one-step diffusion framework for streaming VSR with locality-constrained sparse attention and a tiny conditional decoder.
davidzha712/MoneyPrinterTurbo-H3
Novel-to-video orchestration with local ComfyUI MiniMax H3, Ref2VA and secure Docker defaults
davidzha712/ptv3-pure-blackwell
Pure-PyTorch Point Transformer V3 (serialized attention + U-Net) that trains on Blackwell RTX 50-series (sm_120) — no spconv/torch_scatter. Includes the debugged sm_120 op blacklist.
davidzha712/ip-as-logo-skill
A compact Agent Skill for highly simplified, rounded, subtly neo-skeuomorphic IP mascot logos.
davidzha712/GeoLibre
A lightweight, cloud-native GIS platform for visualizing, exploring, and analyzing geospatial data. It runs in the web browser, on the desktop, on mobile, and inside Jupyter notebooks.
davidzha712/spconv-blackwell-cu128
Verified from-source build of spconv + cumm for CUDA 12.8 / Blackwell sm_120 (RTX 50-series) — forward + backward, unblocks PTv3
davidzha712/claude-code-hud
A fast two-line Claude Code status HUD with Opus Plan routing, context, cost, Git, cache, and rate-limit telemetry.
davidzha712/ai-website-cloner-template
Clone any website with one command using AI coding agents
davidzha712/vllm-dflash-budget-gb10
vLLM v0.21 + DFlash + thinking_token_budget for Gemma 4 & Qwen 3.6 on Blackwell GB10 (sm_121a / sm_120)
davidzha712/vllm
A high-throughput and memory-efficient inference and serving engine for LLMs
davidzha712/goClone
🌱 goClone - clone websites in seconds
davidzha712/pixel-perfect-cloner
Enhanced AI website-cloning workflow: extraction audit + parity completeness gate + 25 field lessons for true 1:1 clones
davidzha712/vllm-qwen3-dflash-budget
vLLM v0.21 + Qwen 3.6 DFlash + real thinking_token_budget enforcement on Blackwell (sm_120 / sm_121a)
davidzha712/matterport
Immersive digital twin platform — Matterport 3D spaces with AI-powered object recognition and asset management
davidzha712/knowflow-portal
KnowFlow AI Vendor Portal - Landing Page + License Management
davidzha712/andreas-magdanz-portfolio
Photography portfolio for Andreas Magdanz — Next.js 16, Sanity v5, GSAP, Tailwind CSS 4
davidzha712/grafana-railway
Deploy Grafana OSS to Railway with one click — PostgreSQL auto-provisioned
davidzha712/tavily-key-generator
Auto batch register Tavily API Keys with pluggable email backends
davidzha712/digital-poststelle
Multimodal NLP email classification system for the construction industry. Published at CIB W78 2024.
davidzha712/davidzha712.github.io
Personal portfolio — Senior ML Engineer & Data Scientist
davidzha712/superset-railway
Deploy Apache Superset 4.x to Railway with one click — connect 50+ databases
davidzha712/tweet-generator
davidzha712/Step-Audio
davidzha712/Code-challenge
davidzha712/davidzha712
My GitHub profile