I'm an AI & Machine Learning Engineer specializing in building impactful solutions for the healthcare industry. My passion is leveraging cutting-edge NLP and deep learning techniques to improve patient outcomes.
- 🧠 Currently working on: Refining a clinical risk prediction model and developing NLP_multihead_multiclass, a multi-task learning framework for DICOM text utilizing BioBERT and parameter-efficient fine-tuning.
- 🌱 Currently learning: Advanced MLOps practices, multi-cloud deployment (AWS), and exploring cutting-edge quantization techniques for LLMs.
- 📫 How to reach me: linkedin.com/in/shiva-heydari/
- NLP_multihead_multiclass: A robust PyTorch-based NLP framework tailored for the healthcare domain. It features multi-head, multi-class classification on DICOM text data leveraging models like BioBERT. It implements advanced, memory-efficient training strategies using Hugging Face PEFT (LoRA) and QLoRA with 4-bit/8-bit quantization.
| Category | Skills |
|---|---|
| Languages | Python, SQL |
| ML/AI Frameworks | PyTorch, Scikit-learn, TensorFlow, NumPy, Pandas, spaCy |
| NLP & LLMs | Hugging Face Transformers, PEFT (LoRA/QLoRA), BioBERT, LangChain, LlamaIndex, RAG, Vector Databases |
| Cloud & MLOps | AWS (SageMaker, ECS, EKS), Docker, Kubernetes, Terraform, CI/CD |
| Databases | MySQL, MongoDB |
| Software Tools | Git, Jupyter |
I'm passionate about the intersection of AI and clinical practice and I'm always open to connecting with fellow researchers, engineers, and pioneers in the health tech space.