Solo ML engineer building high-performance AI agents. Focused on real-world applications where on-device inference and agentic reasoning converge.
Drop a 3-hour video. Get a timestamped intelligence dossier with speaker claims, topic maps, and highlight clips. Powered by Qwen3-VL-32B-Thinking + VibeVoice ASR on AMD MI300X at 0.15× real-time.