Santiago Valdarrama
Founder, Underfitted | AI Educator, Machine Learning Engineer & Content Creator
Founder of Underfitted and leading machine learning educator with over 400,000 followers on X and YouTube. Veteran software architect and machine learning engineer who bridges the gap between complex theoretical research papers and production-ready Python code. Renowned for visual masterclasses on embeddings, vector search, LLM fine-tuning, retrieval architectures, and real-world system design.
Areas of focus
Professional niches
Proof of Work
Hands-on Fine-Tuning Open Source LLMs for Production Tasks
Created open-source walkthroughs showing practitioners how to prepare structured JSONL datasets, apply LoRA adapters via Hugging Face PEFT, and quantize models using bitsandbytes on a single GPU.
Single-GPU fine-tuning restricts effective batch sizes without gradient accumulation and activation checkpointing.
Context: Llama 3 8B, Mistral 7B, PyTorch, Hugging Face TRL and PEFT.
View missionFrom Scratch: Intuitive Vector Search & Embedding Spaces
Authored an interactive architectural breakdown explaining how high-dimensional vectors, dot-product projections, cosine similarities, and Hierarchical Navigable Small World (HNSW) graphs work under the hood.
HNSW approximate nearest neighbors algorithms trade off index build time and RAM consumption for query latency.
Context: Python, NumPy, FAISS, and Chroma vector database architectures.
View mission