New features:
- SerialManager for USB/Serial communication with hardware
- Support for 4 physical VU meter dials
- Flexible channel mapping: Audio L/R, Peak, Mono, CPU, RAM, Disk, Network
- Multiple protocols: Raw bytes, Text, JSON, VU-Server compatible
- Per-dial configuration: min/max values, inversion, smoothing
- Hardware panel in main view showing dial status
- Hardware settings sheet for configuration
- Auto-detection of USB serial devices
Protocol formats:
- Raw: [0xAA][D1][D2][D3][D4][0x55]
- Text: CH1:val;CH2:val;CH3:val;CH4:val\n
- JSON: {"dials":[d1,d2,d3,d4]}
- VU-Server: #0:val\n#1:val\n...
Features:
- Real-time audio level monitoring via BlackHole virtual audio device
- Classic VU meter display with dB scale (-60 to 0 dB)
- Peak hold indicators with configurable hold time
- System resource monitors: CPU, RAM, Disk, Network
- SwiftUI interface with dark theme
- Multi-device audio input selection
- Settings window for configuration
Built with AVAudioEngine for audio capture and Mach kernel APIs
for system statistics.
A Python CLI tool for generating financial reports from Paperless-ngx:
- Phase 1 (MVP): Config handling, Paperless API client with auth and
pagination, custom fields extraction, tag-based summation, CLI output
- Phase 2 (Grouping): Multiple grouping criteria (tag, correspondent,
category, payment type, month, quarter, year), percentage distribution
- Phase 3 (Reports): HTML reports with Chart.js diagrams (doughnut, bar,
line charts), PDF export via WeasyPrint, JSON and CSV export
- Phase 4 (Comfort): Automatic tag ID resolution, disk caching with
diskcache, colorized logging, comprehensive error handling
Features:
- Flexible date filtering (year, month, date range)
- Period comparison with change analysis
- Swiss franc formatting (CHF with apostrophe separators)
- Interactive HTML reports with sortable tables and document links
- Multiple output formats (CLI, HTML, PDF, JSON, CSV)
Complete implementation of automated GitHub repository synchronization:
- Webhook-based auto-sync from GitHub
- Multi-repository support with branch selection
- Web dashboard for management
- Manual sync and rollback functionality
- Comprehensive logging and monitoring
Located in /gitpusher/ subdirectory as standalone application.
Implemented a full-stack web application for fine-tuning LLMs on email data, optimized for Apple Silicon (M4 Pro with 24GB RAM).
Features:
- Mail import with drag & drop support (.mbox, .eml, .txt)
- Automated mail cleaning and preprocessing
- Interactive labeling interface with keyboard shortcuts
- Training data export to JSONL format
- MLX-based LoRA fine-tuning with live updates
- Model evaluation and comparison interface
- Server-Sent Events for real-time training progress
- Dark theme UI optimized for extended use
Technical Stack:
- Backend: FastAPI with SQLite database
- Frontend: Vanilla HTML/CSS/JavaScript (no external dependencies)
- ML Framework: MLX for Apple Silicon optimization
- Models: Support for Mistral 7B and Llama 3 8B via MLX
Components:
- data_manager.py: SQLite operations for mail storage and labeling
- mail_parser.py: Parser for multiple mail formats with cleaning
- training.py: MLX training wrapper with LoRA support
- inference.py: Model loading and inference for evaluation
- main.py: FastAPI backend with REST API and SSE
- Frontend: Complete UI with all features
Documentation:
- Comprehensive README with installation and usage guide
- Quick-start guide for rapid setup
- Example mails for testing
- Troubleshooting and best practices
Ready for local deployment and fine-tuning workflows.
Claude
admin