20 years building system software (Linux kernel, Android TV, Widevine DRM, IPTV). Now applying that systems-level rigor to ML engineering: fine-tuning, RAG pipelines, recommendation systems, and deploying models in production environments.
- 🎬 Content RecSys (two-tower, collaborative filtering, embeddings) for IPTV/OTT movies, series & EPG
- 🏠 Smart home AI assistant (on-device inference, latency-optimized)
- 📡 Fine-tuned 14B LLM (local inference) for MikroTik network management (config, diagnostics, automation)
- 👁️ Test automation combining Computer Vision (YOLO, SAM 3) with LLM log analysis
- 🖨️ 3D printing business pipeline — filament procurement → production → Ozon marketplace (unit economics automation)
ML/AI: PyTorch · HuggingFace · YOLO · SAM 3 · ONNX · llama.cpp
Systems: C/C++ · Python · Linux Kernel · Android AOSP
STB/DTV: Cobalt · WebKit · CI/CD · Widevine DRM · Android TV
- CWIP — Certified Widevine Implementation Partner
- Widevine 3PL Lab — Device certification for Widevine on CE and AOSP
- Android TV Certified — First-ever deployed Linux STB fleet migration to ATV Operator Tier
- Linux Kernel Mainline Contributor
- MBA — MIRBIS, AMBA Accredited
Helping hardware/embedded/Pay-TV companies integrate AI and ML — fine-tuning, RAG, on-device inference, and full pipeline deployment.