Lukasz Gawenda
Lukasz Gawenda shares guides for deploying AI models like Qwen, Gemma, and Llama on local hardware.
About Lukasz Gawenda
Lukasz Gawenda is a Lead Gen AI Engineer who creates content focused on running powerful generative AI models locally and optimizing them for performance. The channel covers LLM and VLM optimization using tools like vLLM configs and KV caching, local AI deployment via Docker, and building custom multimodal workflows with video pre-processing. Gawenda states the channel shares the unfiltered reality of working in generative AI with hands-on hardware-level engineering.
The content includes practical benchmarks and comparisons of models like Qwen 3.8, Gemma 4, and DeepSeek V4, testing speed optimizations such as DFlash, n-gram, SGLang, and quantization techniques. Recent videos feature performance comparisons between different model variants and optimization methods, demonstrating real-world speed improvements and deployment strategies on various hardware configurations.