Skip to content
View AnsKM's full-sized avatar

Highlights

  • Pro

Block or report AnsKM

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
AnsKM/README.md

Ans Khalid

AI Engineer | Full-Stack Developer | Systems Architect

Building production-ready AI systems that deliver measurable business impact. Specializing in real-time audio processing, LLM integration, and desktop applications.

Munich, Germany 🇩🇪 | German (C1) | Open to AI/Backend roles

📧 ansmohal@gmail.com | LinkedIn | 💼 Portfolio


🎯 Core Technical Expertise

AI & LLM Integration

Python Gemini LangChain

Impact: Built 5+ production AI systems serving 100+ users

  • Meeting Assistant: Real-time AI suggestions with <2s response time
  • Memory Agent: Sub-100ms vector search with Qdrant + Voyage AI
  • Claude Workflows: 6-module automation suite for content processing

Key Technologies: GPT-5.1, Claude Opus 4.5, Gemini 3.0, RAG pipelines, Vector DBs (Qdrant, ChromaDB, Pinecone)


Voice AI & Real-Time Audio Processing

Rust Whisper Deepgram

Impact: Processed 100+ hours of voice data with <300ms latency

  • Whisper-Flow: Cross-platform voice dictation with local/cloud options
  • Meeting Assistant: Dual-channel audio capture via Core Audio Taps
  • Real-time STT: Multi-provider architecture (Deepgram, Whisper, Gemini Live)

Key Technologies: Tauri, CPAL, Deepgram Nova-3, Whisper.rs, tokio async runtime


Desktop Application Development

Tauri TypeScript Electron

Impact: Shipped 3 cross-platform desktop apps with 20K+ LOC

  • Meeting Assistant: 20,000+ lines of Rust/TypeScript production code
  • Free Cluely: Invisible AI assistant with local/cloud LLM support
  • Whisper-Flow: Voice dictation with GPU acceleration (macOS/Windows/Linux)

Key Technologies: React 19, Zustand, SQLite, native system integration (Core Audio, IPC)


📦 Full-Stack Development (Click to expand)

Frontend

  • React + TypeScript: Production apps with Vite, Material-UI, Tailwind CSS
  • State Management: Zustand, TanStack Query, Redux patterns
  • Build Tools: Vite 7, Webpack, optimized for performance

Backend

  • Python: FastAPI, Flask (MCP servers, AI APIs)
  • Node.js: Express, RESTful APIs, WebSocket servers
  • Rust: Tauri backends, system-level programming

Databases

  • Vector DBs: Qdrant, ChromaDB, Pinecone (semantic search)
  • Relational: PostgreSQL, SQLite (local-first architecture)
  • NoSQL: Firebase Firestore, Redis caching
🤖 AI/ML Ecosystem

LLM Providers

  • Google Gemini 2.5/3.0, OpenAI GPT-5.1, Claude 4.5, Grok 4.1, DeepSeek, Mistral

Frameworks & Tools

  • LangChain, LangGraph (multi-agent systems), LlamaIndex
  • Prompt engineering, context optimization, RAG architectures
  • Model Context Protocol (MCP) server development

Voice AI Stack

  • Whisper (local inference with CPU/GPU)
  • Deepgram (streaming transcription)
  • Groq API (fast inference)
🛠️ DevOps & Infrastructure
  • Containerization: Docker, multi-stage builds
  • CI/CD: GitHub Actions, automated testing
  • Cloud: Azure, Firebase (Auth, Functions, Firestore)
  • Automation: n8n workflows, Apify actors, web scraping

🔧 Additional Technologies

Systems Programming: Core Audio Taps API, CPAL, audio processing, IPC Web: Chrome Extension APIs (Manifest v3), Shadow DOM, content scripts Tools: Git, npm/cargo/pip, Turborepo monorepos, Firebase CLI Languages: SQL, Bash/Shell, HTML/CSS


🎙️ Meeting Assistant - Real-Time AI Meeting Copilot

Status Private Lines

Private Repository | 20,000+ lines of Rust/TypeScript | Production MVP

A sophisticated desktop application providing real-time AI assistance during technical meetings through live audio processing and intelligent response generation.

Business Impact

  • Sub-2 second AI response generation with context-aware prompting
  • <300ms latency real-time transcription with Deepgram Nova-3
  • Zero driver installation via macOS Core Audio Taps API (14.2+)
  • 90% cost reduction with explicit prompt caching

Technical Architecture

Audio Pipeline:

[Microphone + System Audio] → Core Audio Taps API
  ↓
24kHz Resampling (Rubato) → VAD Filter → Crossbeam Channels
  ↓
Multi-STT (Deepgram/Gemini Live/Whisper) → SQLite Storage
  ↓
Context Assembly → Gemini 2.5 Flash Lite → Streaming UI (SSE)

Key Technical Achievements

  • Dual-Channel Audio Capture: Simultaneous mic + system audio via native macOS API
  • Multi-Provider STT: Hot-swappable between Deepgram, Gemini Live, OpenAI Whisper
  • Advanced Audio Processing: 24kHz resampling with anti-aliasing, VAD filtering
  • Async Architecture: Tokio runtime with lock-free crossbeam channels
  • Local Persistence: SQLite with UUID v7 time-ordered session IDs
  • Context Management: RAG-style meeting prep document loading

Tech Stack

  • Backend: Rust, Tauri 2.0, CPAL, tokio, rubato, whisper-rs
  • Frontend: React 19, TypeScript, Zustand, TanStack Query, Vite
  • AI/STT: Gemini Live, Deepgram Nova-3, OpenAI Whisper
  • Database: SQLite with sqlx, async migrations

Features

  • 🎤 Real-time dual audio capture (no drivers needed)
  • 🧠 Context-aware AI with loaded meeting prep notes
  • ⚡ Sub-500ms end-to-end latency
  • 🔒 100% local data storage (no cloud sync)
  • 🎨 Transparency mode with adjustable opacity
  • ⌨️ Global keyboard shortcuts

[📸 Screenshots Coming Soon] | [📄 Architecture Documentation Coming Soon]

Cost: <$1.00 per 2-hour meeting session


🎤 Murmur - AI Voice Dictation with Privacy

License Platform

View on GitHub | Tauri + React + Rust | Cross-Platform

AI-powered voice dictation desktop app with local-first privacy and flexible backend options.

Business Impact

  • 100% privacy option via local Whisper inference
  • Multi-language UI support (English, Spanish, French)
  • Cross-platform deployment (macOS, Windows, Linux)
  • Auto-learning dictionary with AI-powered suggestions

Technical Highlights

  • Local Whisper Inference: CPU and GPU-accelerated (Vulkan/Metal)
  • Streaming Transcription: Deepgram Nova-3 WebSocket integration
  • LLM Post-Processing: Custom tone personalization (Groq Llama)
  • Smart Dictionary: Glossary terms + replacement rules + auto-learning
  • Turborepo Monorepo: Shared packages for voice-ai, types, utilities

Architecture Philosophy

"Rust is the API, TypeScript is the Brain"

  • Rust provides native capabilities (audio, keyboard, SQLite, Whisper)
  • TypeScript handles all business logic and state management
  • Repository pattern abstracts local (SQLite) vs. cloud (Firebase)

Tech Stack

  • Framework: Tauri 2, React 19, TypeScript, Vite 7
  • State: Zustand with Immer
  • Audio: CPAL, whisper-rs, WebSocket (tokio-tungstenite)
  • AI: Whisper (local), Groq API, Deepgram Streaming
  • UI: Material-UI v7, React Router v6, react-intl

Download Releases | View Documentation


🤖 Claude Code Workflows - AI Content Automation Suite

Status Modules

View on GitHub | Python + FastAPI + Gemini | 6 Specialized Modules

Comprehensive AI-powered tools for content extraction, market analysis, and automation.

Business Impact

  • 6 production modules serving multiple business use cases
  • Automated workflows saving hours of manual research
  • Market intelligence for competitive analysis and strategy
  • Educational content generation from YouTube videos

Modules

  1. YouTube Educator

    • Converts videos → study guides, lesson plans, references
    • AI-generated summaries and code snippet extraction
  2. Meta Ad Analyzer

    • Facebook/Meta ad scraping via Apify
    • Competitive intelligence and performance tracking
    • Export to JSON/CSV/Excel with web dashboard
  3. YouTube Trends

    • Multi-channel monitoring and trend detection
    • Viral prediction scoring with PostgreSQL/SQLite storage
  4. AI Offers Intelligence

    • Google Sheets integration for market analysis
    • Interactive dashboards with price trend analysis
  5. AI Job Hunter

    • Real job scraping from StepStone.de, Indeed.de (Apify)
    • Flask dashboard with salary trends and skills analysis
  6. LinkedIn Personal Branding

    • Complete profile scraping and strength assessment
    • 7/30/90-day actionable growth plans

Tech Stack

  • AI: Google Gemini 2.5, prompt engineering
  • Scraping: Apify actors, BeautifulSoup
  • Web: FastAPI, Flask dashboards
  • Storage: Google Sheets API, PostgreSQL, JSON/CSV
  • Tools: Batch processing, automated reporting

View Tools Documentation | Download Channel Transcripts Tool


🖥️ More Projects

Desktop Applications & Browser Tools

Free Cluely - Invisible AI Assistant

Electron + React + TypeScript | Cross-Platform

Translucent desktop app for real-time AI assistance during meetings

  • Dual AI Modes: Cloud (Gemini 2.5) or Local (Ollama)
  • Screenshot Analysis: Smart content analysis with vision models
  • Global Hotkeys: System-wide shortcuts for instant access
  • Privacy-First: Auto-delete screenshots, local processing option

Tech: Electron, React, Gemini API, Ollama, FFmpeg View on GitHub | Download


Memory Agent MCP Server

FastAPI + Qdrant + MCP Protocol | AI Infrastructure

Persistent memory system for Claude Code with semantic search

  • Sub-100ms retrieval via vector embeddings (Voyage AI)
  • Project isolation with separate memory stores
  • Smart ranking: Relevance + recency + importance
  • Cost-effective: ~$12/month VPS hosting

Tech: Python, FastAPI, Qdrant, Voyage AI, MCP View on GitHub


AI Form Filler - Chrome Extension

JavaScript + Chrome Extension API | Shadow DOM

One-click job application auto-fill with intelligent field detection

  • German + English field name recognition
  • 30+ field types: Personal, professional, education, job-specific
  • Domain filtering: Job site detection and smart activation
  • Privacy-focused: All data stored locally

Tech: Chrome Manifest v3, Shadow DOM, content scripts Status: Production-ready

Enterprise & SaaS Projects

Zeit Erfassung System

Node.js + React + PostgreSQL | 100+ Users

Enterprise time-tracking system with AI-powered features

  • 40% efficiency improvement in time management
  • DORA-compliant architecture for regulated environments
  • German labor law compliance (Arbeitszeitgesetz)
  • Redis caching for performance at scale

Tech: Node.js, Express, React 18, TypeScript, PostgreSQL 16, Redis, Docker


Cold Email AI System

LangGraph + Gemini + Apify | B2B SaaS

Multi-agent automation for deeply-personalized outreach

  • LangGraph orchestration for agent workflows
  • Apify integration for LinkedIn/web scraping
  • 5+ active clients with recurring revenue
  • MCP tool integration for campaign state management

Tech: Python, LangGraph, Gemini API, Apify, MCP


Time Tracker (PentaSoft)

React + Node.js + PostgreSQL | Enterprise

Employee time tracking for 100+ employees

  • Two-layer timesheet system (daily + weekly)
  • Dashboard & reporting with export capabilities
  • German business hours compliance

Tech: React, Node.js, PostgreSQL

Tools & Utilities

Resume/Cover Letter Generation

Python + WeasyPrint + BeautifulSoup

Professional German-style document generation from markdown

  • DIN 5008 compliance for business letters
  • Photo processing for German CV format
  • Structure validation with age-based bullet limits
  • n8n workflow integration with Gemini 2.0 Flash

YouTube Channel Transcript Downloader

Python + yt-dlp + YouTube API

Bulk transcript download for entire YouTube channels

  • Smart error handling with yt-dlp fallback
  • Multi-language and format support (TXT/JSON)
  • Resume capability (skip existing transcripts)
  • Organized storage with metadata tracking

Job Description Extractor

Chrome Extension + Content Scripts

Automatically extract and structure job posting data

  • LinkedIn/Indeed scraping
  • Google Sheets integration
  • Structured export (JSON, CSV)

📊 GitHub Activity

GitHub Stats

GitHub Streak

Highlights

  • 🔨 14+ Active Repositories (public + private)
  • 20,000+ Lines of production Rust/TypeScript in Meeting Assistant
  • 🌟 6 Production Systems deployed and maintained
  • 🤝 Open Source: MCP servers, desktop apps, automation tools

🗺️ Technology-to-Project Mapping

Quick reference for recruiters scanning for specific technologies:

Technology Primary Projects Business Impact
Rust Meeting Assistant, Whisper-Flow Real-time audio processing, system programming
Python Memory Agent, Claude Workflows, AI Job Hunter AI/ML pipelines, automation, web scraping
TypeScript Meeting Assistant, Free Cluely, Whisper-Flow Desktop apps, React frontends, type safety
Tauri Meeting Assistant, Whisper-Flow Cross-platform desktop with native performance
LLM APIs All AI projects Production AI integrations across 6+ systems
Voice AI Meeting Assistant, Whisper-Flow 100+ hours processed, <300ms latency
Vector DBs Memory Agent, RAG systems Sub-100ms semantic search, context retrieval
FastAPI Memory Agent, AI APIs High-performance Python backends
React All desktop apps, Zeit Erfassung Modern UI with hooks, state management
n8n Resume tailor, automation workflows No-code integration, business automation

💼 Open to Opportunities

I'm actively seeking AI Engineer, Full-Stack Developer, or Backend Engineer roles in Munich, Germany or remote within the EU.

What I Bring

  • Production AI experience: Built 5+ systems serving 100+ users with measurable ROI
  • Systems programming: Real-time audio processing, desktop apps, performance optimization
  • Full-stack capabilities: React/TypeScript frontends, Python/Rust backends, vector DBs
  • Business impact focus: 40% efficiency gains, <$1 per session costs, scalable architectures
  • German market fit: C1 fluency, regulatory awareness (DORA, GDPR), DACH experience

Let's Connect

Interested in: AI/ML Engineering, Technical Leadership, Full-Stack Development with AI integration


"Building AI systems that deliver business value while maintaining trust—especially in environments where compliance isn't optional."

Pinned Loading

  1. AnsKM AnsKM Public

    AI Engineer | Full-Stack Developer | Systems Architect