Building production AI systems across LLM infrastructure, RAG pipelines, multi-agent architectures, and speech AI, with focus on reliability, performance, and real-world deployment.
Context compression infrastructure for LLM pipelines.
Focused on:
- token reduction
- long-context optimization
- inference efficiency
- latency reduction
- scalable AI workflows
OCR + summarization system for large document collections.
Features:
- OCR ingestion
- PDF/TXT processing
- RAG workflows
- local summarization
- Rust backend architecture
🔗 https://github.com/rizgan/FileRAG
Speech recognition infrastructure using Whisper and VOSK.
Focus:
- multilingual ASR
- speech pipelines
- audio preprocessing
- streaming transcription systems
Machine learning experiments and forecasting systems.
🔗 https://www.youtube.com/@PricePredictionAI