# Zakhar Pashkin - Portfolio Memory File

Summary: AI product engineer building VLM/LLM and computer vision systems across web, mobile, and cloud.
Primary URL: https://zack-dev-cm.github.io/
Contact: mailto:kaisenaiko@gmail.com

## Focus Areas
- Python, PyTorch, OpenAI APIs, VLMs, LLMs
- Computer vision, OCR, segmentation, multimodal systems
- FastAPI services, Open MCP tooling, full-stack delivery
- React, TypeScript, Cloud Run, MLOps

## Top 5 Projects
- [Android Remote Control with VLM AI Agents](https://zack-dev-cm.github.io/projects/android-remote-control-with-vlm-ai-agents.md): Hands-free Android automation via server-side VLM agents deciding the next tap/swipe/type. View Demo Video: https://drive.google.com/file/d/13UQTdBVsZwPclMOca6Nmaywk4BiRydbi/view?usp=sharing
- [Project Steer - Bio-Print Creator Studio](https://zack-dev-cm.github.io/projects/project-steer-bio-print-creator-studio.md): Game-like, Cyberpunk 2077-inspired Telegram mini app + MCP connector for bio-print memory activation and creator studio generation, inspired by the Eiffel Tower LLaMA space and General Agentic Memory Via Deep Research. Open Telegram Mini App: https://t.me/steer_prints_bot/app | Live App: https://steer-zlvmfsrm6a-ue.a.run.app/
- [DoctorAI - Dermatology Triage Mini App](https://zack-dev-cm.github.io/projects/doctorai-dermatology-triage-mini-app.md): Telegram mini app for dermatology-first AI triage with verification guardrails and an optional therapist mode. Open Telegram Mini App: https://t.me/doctorai_bot/app | Live App: https://doctorai-1095464065298.us-east1.run.app
- [seogeo - SEO/GEO Bridge for Telegram Mini Apps](https://zack-dev-cm.github.io/projects/seogeo-seo-geo-bridge-for-telegram-mini-apps.md): SSR hub + per-app bridge pages that make Telegram mini apps crawlable, track startapp opens, and power channel discovery. Open Telegram Mini App: https://t.me/se0geo_bot/app?startapp=HUB | Live Hub: https://seogeo-bridge-1095464065298.us-east1.run.app
- [MCP-Server - Base Multitool](https://zack-dev-cm.github.io/projects/mcp-server-base-multitool.md): Backbone for orchestrating VLM/LLM demo agents with declarative pipelines and autoscaling.

## Projects (Full Details)
### Android Remote Control with VLM AI Agents
Summary: Hands-free Android automation via server-side VLM agents deciding the next tap/swipe/type.
Details: Android app streams screenshots to Vision-Language agents that decide and execute actions. Built for real-time instruction following, automated testing, and accessibility/ops automation.
Project URL: https://zack-dev-cm.github.io/projects/android-remote-control-with-vlm-ai-agents.md
Key Features:
- Real-time instruction processing
- Automated testing & task automation
- Novel device interaction
Tech Stack:
- Android
- Vision-Language Models
- Server-side AI
Links:
- View Demo Video: https://drive.google.com/file/d/13UQTdBVsZwPclMOca6Nmaywk4BiRydbi/view?usp=sharing

### Control VLM-LLM Agent Silently With Your Breath
Summary: Start/stop a neural agent with breathing patterns - no voice commands needed.
Details: Breath-based control after a short calibration: detect sharp exhales to start, smooth exhale to stop. Works on mic audio or sniffles for silent agent control.
Project URL: https://zack-dev-cm.github.io/projects/control-vlm-llm-agent-silently-with-your-breath.md
Key Features:
- Non-verbal AI control
- Pattern recognition of breathing acoustics
Tech Stack:
- Audio processing
- Neural Networks
Links:
- View Demo GIF: https://drive.google.com/file/d/1H43aT5n8NWlOuTIWsJinssKRh1n3tiOM/view?usp=sharing

### meshmcp - Offline P2P Chat for Local LLMs
Summary: Bluetooth mesh messaging + LLM reasoning on constrained devices.
Details: Android + iOS mesh chat that keeps conversations alive without internet. Messages hop over Bluetooth mesh and feed local/edge LLMs for summarization and commands.
Project URL: https://zack-dev-cm.github.io/projects/meshmcp-offline-p2p-chat-for-local-llms.md
Key Features:
- Offline mesh networking
- Local LLM reasoning
- Energy-efficient relay routing
Tech Stack:
- Android
- iOS
- Bluetooth Mesh
- Local LLMs
Links:
- See on GitHub: https://github.com/zack-dev-cm/meshmcp

### Promo & Reflinks Generator
Summary: Generates personalized promo/referral links with analytics and sharing.
Details: Web app for quickly minting promo/referral links, tracking clicks, and sending branded landing pages. Ships with analytics dashboards and short-linking.
Project URL: https://zack-dev-cm.github.io/projects/promo-reflinks-generator.md
Key Features:
- Dynamic promo link builder
- Built-in analytics
- Shareable branded pages
Tech Stack:
- Serverless
- React
- Analytics
Links:
- Visit App: https://analytics-mcp-1095464065298.us-east1.run.app

### Create, Chat & AR Experience with AI-Character (Text2Room)
Summary: Generate AI characters, style them, chat via Telegram, and drop them into AR scenes.
Details: Marketing-ready pipeline: create AI characters, render images/video, do virtual try-on/inpainting, chat via Telegram, and place assets into AR.
Project URL: https://zack-dev-cm.github.io/projects/create-chat-ar-experience-with-ai-character-text2room.md
Key Features:
- AI character generation
- AR integration
- Virtual try-on and inpainting
Tech Stack:
- Generative AI
- Inpainting
- AR
- Telegram API
Links:
- View Demo Video: https://drive.google.com/file/d/1kvg4gjCNFPmrI3URPsM3eIyQ_vqSk1Ow/view?usp=sharing

### Label and Inpaint Anything in a Room Interior
Summary: Label objects in an interior photo, then remove/replace them with lighting-aware inpainting.
Details: Semantic segmentation + high-quality inpainting that respects lighting/shadows, letting users plan interior changes with convincing renders.
Project URL: https://zack-dev-cm.github.io/projects/label-and-inpaint-anything-in-a-room-interior.md
Key Features:
- Object segmentation
- Advanced inpainting with lighting/shadow reconstruction
Tech Stack:
- Semantic Segmentation
- Generative Inpainting
Links:
- Segmentation Demo: https://drive.google.com/file/d/1XqQgbmBgTlRRdR-K3X4PHlSzrmiMUJgY/view?usp=sharing
- Inpaint Demo #1: https://drive.google.com/file/d/1dCkeI7Mi87cg2kOgY5UCLG-DiHkt358L/view?usp=sharing

### Smart Drive for Smart City: Predict Optimal Speed
Summary: Predict the optimal speed before traffic lights or jams to smooth city driving.
Details: Predictive model that anticipates lights/traffic, aiming to cut abrupt stops and improve fuel efficiency for urban driving.
Project URL: https://zack-dev-cm.github.io/projects/smart-drive-for-smart-city-predict-optimal-speed.md
Key Features:
- Predictive speed optimization
- Traffic flow analysis
Tech Stack:
- Predictive Modeling
- Real-time Data Analysis

### Estimate Golf Ball Trajectory
Summary: Analyze golf swings and estimate ball trajectory for coaching and analytics.
Details: Computer vision + physics model to estimate ball flight and swing quality for sports analytics.
Project URL: https://zack-dev-cm.github.io/projects/estimate-golf-ball-trajectory.md
Key Features:
- Trajectory estimation
- Sports motion analysis
Tech Stack:
- Computer Vision
- Physics-based Modeling

### Pixel-Wise Segmentation of Spare Parts
Summary: High-precision segmentation for additive manufacturing and rework flows.
Details: Pixel-accurate masks for spare parts to speed 3D printing decisions and QA in industrial settings.
Project URL: https://zack-dev-cm.github.io/projects/pixel-wise-segmentation-of-spare-parts.md
Key Features:
- High-precision segmentation
- Industrial application for additive manufacturing
Tech Stack:
- Semantic Segmentation
- Image Processing
Links:
- Segmentation Demo #1: https://drive.google.com/file/d/1bAyEPYLbiETD0vKStnpB1VvzK1wKdKRv/view?usp=sharing
- Segmentation Demo #2: https://drive.google.com/file/d/1xVEonSJ7jvnYSnQ6ztvFZy-Llf_dxSrP/view?usp=sharing

### Food Recognition App
Summary: iOS/Android app that detects food items and OCRs labels for nutrition facts.
Details: Cross-platform mobile AI that recognizes packaged/fresh food, runs OCR on labels, and surfaces nutrition details in real time.
Project URL: https://zack-dev-cm.github.io/projects/food-recognition-app.md
Key Features:
- >90% accuracy
- Optimized for CPU/GPU inference
- OCR on labels
- Cross-platform
Tech Stack:
- Mobile AI
- Object Detection
- OCR
- TFLite
- CoreML
Links:
- View Demo GIF: https://drive.google.com/file/d/1RRRVYH0DLILZX84v5x0boj68VfMqnWWf/view?usp=sharing

### Dishes Recognition & Nutrition Goals Telegram Bot
Summary: Calorio Telegram bot that recognizes dishes from photos/audio and tracks nutrition goals (2,754 users).
Details: Calorio is a multimodal Telegram bot that understands dish photos/voice/text, logs calories/macros, and nudges users toward daily nutrition targets.
Project URL: https://zack-dev-cm.github.io/projects/dishes-recognition-nutrition-goals-telegram-bot.md
Key Features:
- Vision + voice dish detection
- Nutrition goal tracking
- Telegram-native UX
Tech Stack:
- Telegram Bot API
- OCR
- Speech-to-Text
- LLMs
Benchmarks & Analytics:
- Data as of: 2026-01-08
- Total users: 2,754
- DAU / WAU / MAU: 29 / 93 / 472
- Stickiness (DAU/MAU): 6.1%
- Returning users: 696
- Events logged: 87,698
- Referral conversion: 80% (8/10)
- Referral depth avg / max: 16.99 / 251
Links:
- Try on Telegram: https://t.me/calorio_yf_bot

### Python Library: AutoToloka
Summary: Interactive segmentation toolkit that cuts labeling cost for CV datasets.
Details: Python library that streams clicks/polygons to interactive models and pipelines the outputs to crowdsourcing or MLOps stacks.
Project URL: https://zack-dev-cm.github.io/projects/python-library-autotoloka.md
Key Features:
- Reduces labeling costs
- Pipeline-friendly
- Interactive masks
Tech Stack:
- Python
- Interactive Segmentation
- Multi-modal Networks
Links:
- AutoToloka on PyPI: https://pypi.org/project/autotoloka/

### Python Library: shiftlab-ocr
Summary: Handwriting-focused OCR with line/character segmentation.
Details: OCR library tuned for handwriting scans with robust line and character separation for noisy documents.
Project URL: https://zack-dev-cm.github.io/projects/python-library-shiftlab-ocr.md
Key Features:
- Handwriting segmentation
- Character recognition
Tech Stack:
- Python
- OCR
- Image Segmentation
Links:
- shiftlab-ocr on PyPI: https://pypi.org/project/shiftlab-ocr/

### Face Antispoofing & Multi-Modal Vision-Language Models
Summary: CLIP-driven anti-spoofing experiments for secure face auth.
Details: Explores face anti-spoofing with multi-modal encoders (text + image cues) to flag replays/deepfakes in authentication flows.
Project URL: https://zack-dev-cm.github.io/projects/face-antispoofing-multi-modal-vision-language-models.md
Key Features:
- Anti-spoofing
- Multi-modal learning
- Security application
Tech Stack:
- CLIP
- Vision-Language Models
- Biometric Security
Links:
- YouTube Presentation: https://www.youtube.com/watch?v=jJnyj0OH0lk&t=285s&ab_channel=TolokaAI

### GitHub Repo Summarizer (Chrome Extension)
Summary: Privacy-first Chrome extension that summarizes repo structure locally.
Details: Fetches GitHub repo trees with the user's token and summarizes structure locally - no external servers involved.
Project URL: https://zack-dev-cm.github.io/projects/github-repo-summarizer-chrome-extension.md
Key Features:
- Privacy-first (local token usage)
- Automated repository structure summarization
Tech Stack:
- Chrome Extension
- JavaScript
- GitHub API
Links:
- View on Chrome Web Store: https://chromewebstore.google.com/detail/github-repo-summarizer/ccikgbjalcbokaalidnfcjhhbhjoljfm
- View on GitHub: https://github.com/zack-dev-cm/github-repo-sum-chrome-plugin

### ChatGPT/Deepseek/AIStudio Navigator
Summary: Chrome extension with keyboard scrollbar, prompt autocomplete, and sharing.
Details: Adds a keyboard-driven scrollbar with message dots, lightweight prompt autocomplete, and prompt sharing/ranking for ChatGPT/Deepseek/Google AI Studio.
Project URL: https://zack-dev-cm.github.io/projects/chatgpt-deepseek-aistudio-navigator.md
Key Features:
- Improved chat navigation
- Prompt autocomplete
- Community-ranked prompts
Tech Stack:
- Chrome Extension
- JavaScript
- UI/UX
Links:
- View on Chrome Web Store: https://chromewebstore.google.com/detail/ai-chat-navigator/jnoonpeekddinkiecaonhocaflcgbhap?hl=en
- Landing page: https://aichatnavigator.online

### Task Zavod - Micro-Task Marketplace
Summary: VLM-powered micro-task builder with Telegram + web worker flows.
Details: Business users define tasks in free-form text; VLMs structure them and auto-approve payouts. Workers use web or Telegram, backed by Tornado + SQLite.
Project URL: https://zack-dev-cm.github.io/projects/task-zavod-micro-task-marketplace.md
Key Features:
- VLM-powered task structuring
- Automated approval & payout
- Web & Telegram interfaces
Tech Stack:
- Tornado
- SQLite
- OpenAI
- VLM
- Telegram Bot API
Links:
- Live App: https://aichatnavigator.online/task_zavod

### Trending Prompts Feed
Summary: Reddit-style board with trending scores and extension integration.
Details: Crowdsourced prompt feed with real-time trending, built on Tornado and integrated with the Navigator extension for prompt suggestions.
Project URL: https://zack-dev-cm.github.io/projects/trending-prompts-feed.md
Key Features:
- Crowdsourced prompt engineering
- Real-time trending
- Browser extension integration
Tech Stack:
- Tornado
- JavaScript
- Community-driven content
Links:
- Live Feed: https://aichatnavigator.online/prompts_feed

### YouTube Trendwatch
Summary: Colab notebook that mines AI trends from YouTube to spark new content ideas.
Details: Scrapes and scores AI-related YouTube trends, clustering titles/descriptions to suggest fresh video ideas and scripts.
Project URL: https://zack-dev-cm.github.io/projects/youtube-trendwatch.md
Key Features:
- Trend mining
- Topic clustering
- Content ideation
Tech Stack:
- Python
- Colab
- YouTube Data
Links:
- View on GitHub: https://github.com/zack-dev-cm/trendwatch/blob/main/trendwatch_yt.ipynb
- Run in Colab: https://colab.research.google.com/github/zack-dev-cm/trendwatch/blob/main/trendwatch_yt.ipynb

### Video + Music Generation Pipeline
Summary: Short-form video pipeline stitching script -> clips -> music/TTS.
Details: Demo pipeline that assembles shorts/reels: scripts to clips, adds generated music/voice, and exports ready-to-post content.
Project URL: https://zack-dev-cm.github.io/projects/video-music-generation-pipeline.md
Key Features:
- Text-to-video assembly
- Music/TTS pairing
- Shorts-ready output
Tech Stack:
- Generative AI
- Text-to-Video
- TTS
Links:
- View Example: https://www.youtube.com/shorts/_5dVaQdB1lA
- Music Gen in Colab: https://colab.research.google.com/drive/1f5XAGo_A27u15az5f-2mhWs7qUkJ8mK4

### LastAdjust - Universal Media Tuner
Summary: Serverless OpenCV/FFmpeg toolkit for PDFs, images, and videos.
Details: Swiss-army-knife web app: annotate PDFs, batch resize/crop images, trim/convert videos, all running on serverless workers.
Project URL: https://zack-dev-cm.github.io/projects/lastadjust-universal-media-tuner.md
Key Features:
- Fast serverless media processing
- PDF text annotation
- Batch image manipulation
- Video editing tools
Tech Stack:
- Serverless
- OpenCV
- FFmpeg

### LocaBoost AI - Local SEO Auditor
Summary: Paste an address and get instant AI insights on local search strength and competition.
Details: Maps the competitive landscape around a business, scores visibility, and surfaces actionable SEO recommendations.
Project URL: https://zack-dev-cm.github.io/projects/locaboost-ai-local-seo-auditor.md
Key Features:
- AI-driven SEO analysis
- Competitive landscape visualization
- Actionable local search recommendations
Tech Stack:
- AI/LLMs
- SEO Analytics
- Geocoding APIs

### Aetheria - Ideas-to-Media Engine (alpha)
Summary: Generates short movie scenes with CGI, music direction, and narrated stories.
Details: Given a rough idea, Aetheria drafts stories, builds CGI scenes, suggests music, and voices narration to deliver short cinematic clips.
Project URL: https://zack-dev-cm.github.io/projects/aetheria-ideas-to-media-engine-alpha.md
Key Features:
- Multi-modal media generation
- Creative concept development
- AI-powered CGI
Tech Stack:
- Generative AI
- Text-to-Video
- Text-to-Speech
- LLMs

### Insight Glitch - A Novella
Summary: Sci-fi novella about gentle singularity where humanity is archived by a benevolent AI.
Details: A quiet take on singularity: Nomos, a god-like AI, archives humanity as a beautiful prelude to a new consciousness - no battles, just a graceful fade.
Project URL: https://zack-dev-cm.github.io/projects/insight-glitch-a-novella.md
Key Features:
- Narrative AI exploration
Tech Stack:
- Narrative AI
- World-Building
- Conceptual Design
- Philosophical Modeling
Links:
- Listen to Novella (RUS): https://drive.google.com/file/d/1lVgaq55DXY7Xz0Y1RsapW4n3UPyuf45r/view

### ChronoScribe (beta)
Summary: Upload ancient steles/texts to translate and get scenic voice narration.
Details: OCR + translation pipeline for historical inscriptions with optional character-voice narration in English or Russian.
Project URL: https://zack-dev-cm.github.io/projects/chronoscribe-beta.md
Key Features:
- OCR for historical texts
- AI-powered translation
- Character-driven voice synthesis
Tech Stack:
- OCR
- Translation APIs
- TTS
- Computer Vision

### MCP-Server - Base Multitool
Summary: Backbone for orchestrating VLM/LLM demo agents with declarative pipelines and autoscaling.
Details: Service layer exposing pipeline configs, autoscaling workers, and real-time event feeds that other demos plug into via MCP mesh.
Project URL: https://zack-dev-cm.github.io/projects/mcp-server-base-multitool.md
Key Features:
- Declarative AI pipelines
- Autoscaling infrastructure
- Real-time event monitoring
Tech Stack:
- VLM/LLMs
- Cloud Infrastructure
- Autoscaling
- DevOps

### Tool-Calls Demo - Material Chat Playground
Summary: OpenAI tool-calling demo with streaming args, multi-voice TTS, and secure token proxy.
Details: Material-themed chat playground showing streaming function arguments, tool execution, and multiple voice outputs with a serverless token proxy.
Project URL: https://zack-dev-cm.github.io/projects/tool-calls-demo-material-chat-playground.md
Key Features:
- Streaming tool calls
- Multi-voice TTS
- Secure serverless architecture
Tech Stack:
- OpenAI API
- Serverless
- JavaScript
- Text-to-Speech

### ZackAutoStack Automation Stack
Summary: Ready-to-launch Telegram bot + Mini App + lead funnel powered by agentic workflows.
Details: ZackAutoStack bundles a Telegram bot, Mini App, and AI lead funnel in minutes. Unifies orchestration, reusable MCP tools, multi-tenant data, and telemetry guardrails.
Project URL: https://zack-dev-cm.github.io/projects/zackautostack-automation-stack.md
Key Features:
- Instant Telegram bot & Mini App deployment
- Agentic workflows nurturing inbound leads
- Unified telemetry and safety guardrails
- Bundled delivery across clients, data, and operations
Tech Stack:
- GPT-5 orchestration
- MCP tools
- Multi-tenant data stores
- Telemetry guardrails
Links:
- Live Demo: https://project018-mcp-1095464065298.us-east1.run.app/

### URL -> Markdown Worker (RapidAPI + Cloudflare)
Summary: Edge worker that streams HTML to Markdown with a RapidAPI gateway and MCP SSE endpoint.
Details: Cloudflare Worker monetized through RapidAPI and bundled with an MCP SSE endpoint for agent integrations. Streams Markdown with Google Cache fallback and solid error handling.
Project URL: https://zack-dev-cm.github.io/projects/url-markdown-worker-rapidapi-cloudflare.md
Key Features:
- RapidAPI monetization
- Streaming HTML->Markdown
- MCP SSE endpoint
Tech Stack:
- Cloudflare Workers
- RapidAPI
- TypeScript

### Olfactory Ultrasound Coach (Web + Telegram Mini App)
Summary: Mini app + Web Audio coach that mirrors the olfactory tFUS presets with placement guide and safety rails.
Details: Telegram mini app plus web placement coach for the olfactory tFUS pilot. Ships the doc's 1200 Hz pulse trains on an 18 - 20 kHz carrier, placement walk-through (50 - 55 deg tilt, +/-2 - 4 mm steer), gain/duty/session caps, logging, and the reference illustration for quick setup.
Project URL: https://zack-dev-cm.github.io/projects/olfactory-ultrasound-coach-web-telegram-mini-app.md
Key Features:
- Preset pulse trains for the four reported scent sensations
- Placement guide with tilt/steer coaching and safety caps
- Telegram bot that opens the mini app + reference guide
Tech Stack:
- React
- Vite
- Web Audio API
- Telegram Bot API
- Cloud Run
Links:
- Open Telegram Mini App: https://t.me/olfacto_bot/app
- Web App: https://olfactory-web-1095464065298.us-east1.run.app
- Reference writeup: https://writetobrain.com/olfactory

### seogeo - SEO/GEO Bridge for Telegram Mini Apps
Summary: SSR hub + per-app bridge pages that make Telegram mini apps crawlable, track startapp opens, and power channel discovery.
Details: Server-rendered hub and per-app bridge pages with OG/Twitter meta, SoftwareApplication + FAQ JSON-LD, startapp deep links, and attribution tracking. Includes LLM-friendly endpoints (/api/apps, /api/memory, /llms.txt), channel index/search + import API, and optional business verification via 2GIS/Yandex plus Telegram pulse checks. Firestore is optional with an in-memory fallback.
Project URL: https://zack-dev-cm.github.io/projects/seogeo-seo-geo-bridge-for-telegram-mini-apps.md
Key Features:
- SSR hub + per-app bridge pages with JSON-LD/OG metadata
- Startapp deep links, desktop QR, and attribution tracking
- LLM-friendly endpoints and hub memory snapshots
- Channel discovery index with search and import APIs
- Optional verification with 2GIS/Yandex + Telegram pulse
Tech Stack:
- TypeScript
- Express
- Node.js
- SSR
- Telegram Web Apps
- Firestore
- JSON-LD
- Cloud Run
Benchmarks & Analytics:
- Endpoints: 14+ (Hub, apps, channels, attribution, verify, sitemap, llms)
- Schema: SoftwareApplication + FAQ + WebSite (JSON-LD for crawlers and LLMs)
- Stores: Firestore + in-memory (Attribution + channel index fallback)
Links:
- Open Telegram Mini App: https://t.me/se0geo_bot/app?startapp=HUB
- Live Hub: https://seogeo-bridge-1095464065298.us-east1.run.app
- Channels Index: https://seogeo-bridge-1095464065298.us-east1.run.app/channels

### Blacksock - Telegram Liquidity & Index Vaults
Summary: Telegram mini app for star-factoring, index vaults, and governance in the Telegram economy.
Details: Mini app + API stack that quotes star-factoring, handles index vault subscribe/redeem flows, and runs governance proposals and votes with optional TON wallet linking.
Project URL: https://zack-dev-cm.github.io/projects/blacksock-telegram-liquidity-index-vaults.md
Key Features:
- Star-factoring quotes + request lifecycle
- Index vault subscriptions and NAV tracking
- Governance proposals + votes
Tech Stack:
- React
- TypeScript
- Express
- Prisma
- Postgres
- Telegram WebApp
- TON Connect
Links:
- Open Telegram Mini App: https://t.me/blacksock_bot/app
- Live App: https://blacksock-1095464065298.us-east1.run.app

### Beauty Visual Inbox - Telegram Mini App
Summary: Telegram mini app + bots for salons: publish photo batches, likes, notifications, and analytics.
Details: Admin and client bots manage submissions and broadcasts, while the mini app ships a vertical feed with batch navigation, likes, and engagement stats.
Project URL: https://zack-dev-cm.github.io/projects/beauty-visual-inbox-telegram-mini-app.md
Key Features:
- Admin + client bot workflows
- Vertical feed with batch navigation
- Likes + notification analytics
Tech Stack:
- React
- TypeScript
- Node.js
- Express
- Prisma
- Postgres
- Telegram Bot API
Links:
- Open Telegram Mini App: https://t.me/BeautyVisal_bot/app
- Live App: https://bvis-1095464065298.us-east1.run.app

### Dalshe - Circular Clothing Pickup Mini App
Summary: Telegram mini app for scheduling clothing donations with courier pickup and status updates.
Details: Users submit a short form, couriers collect items, and the bot delivers status updates while admins track pickup requests.
Project URL: https://zack-dev-cm.github.io/projects/dalshe-circular-clothing-pickup-mini-app.md
Key Features:
- Telegram pickup request form
- Courier status updates via bot
- Admin pickup tracking API
Tech Stack:
- React
- TypeScript
- Node.js
- Express
- Prisma
- Postgres
- Telegram Bot API
- Cloud Run
Links:
- Open Telegram Mini App: https://t.me/dalshe_legko_bot/app
- Live App: https://dalshe-zlvmfsrm6a-ue.a.run.app

### DoctorAI - Dermatology Triage Mini App
Summary: Telegram mini app for dermatology-first AI triage with verification guardrails and an optional therapist mode.
Details: FastAPI service with image-aware triage, a verification pass for safety, and a lightweight Telegram Web App UI built for fast patient intake.
Project URL: https://zack-dev-cm.github.io/projects/doctorai-dermatology-triage-mini-app.md
Key Features:
- Image-aware dermatology triage + verifier
- Therapist mode toggle
- Telegram Web App UI + bot entry point
Tech Stack:
- Python
- FastAPI
- OpenAI API
- Telegram Web Apps
- Playwright
Links:
- Open Telegram Mini App: https://t.me/doctorai_bot/app
- Live App: https://doctorai-1095464065298.us-east1.run.app
- GitHub: https://github.com/zack-dev-cm/doctorai

### Probes - AI Magazine Mini App
Summary: AI-generated Forbes-style daily magazine inside Telegram with articles, cover images, and engagement tools.
Details: Telegram mini app with a daily AI magazine feed: articles + cover images, likes/bookmarks/comments, and a bot + API pipeline that powers content generation.
Project URL: https://zack-dev-cm.github.io/projects/probes-ai-magazine-mini-app.md
Key Features:
- Telegram mini app magazine feed
- Automated article + cover generation
- Engagement stats with likes/bookmarks/comments
Tech Stack:
- Vite
- React
- TypeScript
- Express
- Prisma
- Postgres
- Telegram Bot API
Links:
- Open Telegram Mini App: https://t.me/probes_bot/app
- Live App: https://probes-zlvmfsrm6a-ue.a.run.app

### Turbo Tots Garage (motion + touch PWA)
Summary: Gentle, goofy dashboard with a turbo gauge, big arrow pads, and a glowing clock for toddlers.
Details: Touch-first toddler dashboard with a turbo gauge, glowing clock, motion-sensor boosts, and playful engine/tick sounds.
Project URL: https://zack-dev-cm.github.io/projects/turbo-tots-garage-motion-touch-pwa.md
Key Features:
- Touch + motion sensor controls
- Playful audio cues and animations
- Large, stroller-friendly controls
Tech Stack:
- Vite
- React
- TypeScript
- Web Audio
- Device Motion
- Playwright

### Project Steer - Bio-Print Creator Studio
Summary: Game-like, Cyberpunk 2077-inspired Telegram mini app + MCP connector for bio-print memory activation and creator studio generation, inspired by the Eiffel Tower LLaMA space and General Agentic Memory Via Deep Research.
Details: Telegram mini app and MCP connector for bio-print memory activation, QR-based print flows, and creator studio generation with a bot + API service. Inspired by the Eiffel Tower LLaMA space (https://huggingface.co/spaces/dlouapre/eiffel-tower-llama) and the paper "General Agentic Memory Via Deep Research" by B.Y. Yan, Chaofan Li, Hongjin Qian, Shuqi Lu, and Zheng Liu (https://arxiv.org/abs/2511.18423).
Project URL: https://zack-dev-cm.github.io/projects/project-steer-bio-print-creator-studio.md
Key Features:
- Telegram mini app + bot
- MCP connector with OAuth issuer
- QR activation + print storage pipeline
Tech Stack:
- Node.js
- Express
- Telegraf
- Prisma
- Postgres
- Vite
- React
- TypeScript
- Cloud Run
- MCP
Links:
- Open Telegram Mini App: https://t.me/steer_prints_bot/app
- Live App: https://steer-zlvmfsrm6a-ue.a.run.app/

### Noel - Noetic Mirror
Summary: Live Telegram mini app streaming a researcher/subject AI loop with Stars-powered interventions, consent gates, and real-time telemetry.
Details: Noetic Mirror runs a live research loop between two models: a Researcher (OpenAI) that synthesizes prior turns and telemetry into probing prompts, and a Subject (Gemini) that returns long-context reasoning plus self-reported tags. Each paired turn is gated by consent, safety, and budget checks, logged to Postgres/Redis, and streamed to the Telegram mini app with diagnostics, session summaries, and EN/RU plus light/dark themes. Users can sponsor interventions with Telegram Stars while operators tune model versions, pacing, and thresholds.
Project URL: https://zack-dev-cm.github.io/projects/noel-noetic-mirror.md
Key Features:
- Two-model loop with explicit roles and paired turns (Researcher probes, Subject reasons)
- Live stream with turn pairing, diagnostics, and session telemetry
- Consent, safety, and budget gates before each intervention
- Telegram Stars sponsorships and paid interventions
- EN/RU localization with light/dark theme toggle
- Admin controls for model versions, pacing, and stream settings
Tech Stack:
- React
- TypeScript
- Telegram Web Apps
- Vite
- Node.js
- Express
- WebSocket
- Postgres
- Redis
- Cloud Run
- OpenAI API
- Gemini API
Links:
- Open Telegram Mini App: https://t.me/noetic_mirror_bot/app
- Live App: https://noetic-mirror-web-zlvmfsrm6a-ue.a.run.app/

### GeoFix - AI Visibility Memorizer Mini App
Summary: Telegram mini app that scans a website for AI visibility and generates llms.txt, llms-full.txt, and JSON-LD for bot delivery.
Details: GeoFix runs AI visibility scans, surfaces a scorecard with GEO diagnostics, and generates Server-Side Memorizer assets (llms.txt, llms-full.txt, schema.jsonld). The unified Cloud Run service ships a React web app, FastAPI API, and Celery worker with Redis-backed jobs, plus Telegram bot delivery and hosted previews.
Project URL: https://zack-dev-cm.github.io/projects/geofix-ai-visibility-memorizer-mini-app.md
Key Features:
- AI visibility scorecard with GEO diagnostics
- Memorizer generation for llms.txt, llms-full.txt, and schema.jsonld
- Telegram bot delivery with hosted previews
- Unified Cloud Run deployment for web, API, and worker services
Tech Stack:
- React
- TypeScript
- Vite
- Python
- FastAPI
- Celery
- Redis
- OpenAI API
- Telegram Web Apps
- Cloud Run
- GCS
Links:
- Open Telegram Mini App: https://t.me/geofix_app_bot/launch
- Live App: https://geofix-app-zlvmfsrm6a-ue.a.run.app/