Articles

How I used ConvNeXt-V2 and MIL attention to classify brain CT scans from 859 real patients — covering dual-head architecture, class imbalance, and lessons from working with clinical data.

A walkthrough of building a production-ready RAG pipeline with FastAPI, pgvector, sentence-transformers, and a local LLM — no cloud API costs required.

How I built a content-based recommendation engine for a microjobs portal using TF-IDF vectorization and Facebook's FAISS for approximate nearest-neighbor search.

How I used LangChain's OpenAI Functions agent with custom tools to build a context-aware writing assistant. Covers tool design, memory integration, and agent orchestration.