๐ Core Technical Interests
LLM Optimization & Quantization: Focus on making large models run efficiently on consumer hardware (GGUF, AWQ, EXL2).
Efficient Fine-Tuning (PEFT): Specialized in LoRA and QLoRA techniques for domain-specific adaptation.
Model Evaluation Frameworks: Building robust pipelines to measure LLM performance beyond basic benchmarks.
On-Device AI: Deploying high-performance models for edge computing and local environments.
๐ Specialized Domains
Algorithmic Trading & Time-Series: Applying Transformers and GRUs to financial markets and predictive signaling.
Agentic Workflows: Designing autonomous AI agents that can navigate complex multi-step tasks.
MLOps & Scalable Infrastructure: Optimizing Dockerized environments for seamless model serving and CI/CD.
โจ Minimalist "Bio" Style
If you prefer a clean, "Next.js-style" aesthetic for your profile, try a bulleted list with emojis:
โก High-Performance Inference (A100/H100 Optimization)
๐ง Fine-Tuning 8B+ Models (Llama, Mistral, XCurOS)
๐ Full-Stack AI Integration (Next.js + FastAPI + Docker)
๐ Neural Time-Series Analysis