AI & Machine Learning
LLM Fine-Tuning, LoRA & Quantization
Custom domain fine-tuning of open weights models (Llama 3, Qwen 2.5, DeepSeek R1) using PEFT/LoRA and GGUF/AWQ quantization.
Core Competency Rating94%
Architecture Blueprint
Reduces model VRAM requirements by 70% while retaining 99.2% benchmark accuracy against full precision base models.
Key Technical Capabilities
Parameter-Efficient Fine-Tuning (LoRA / QLoRA)
Supervised Fine-Tuning (SFT) & DPO / PPO Alignment
Model Compression via AWQ & GGUF Quantization
High-Throughput Inference via vLLM & TensorRT
Tooling & Production Stack
PyTorchUnslothAxolotlDeepSpeedvLLMTensorRT-LLM