Smart Recruiter AI
An agentic, RAG-based resume screener that reads resumes the way a human would instead of doing brittle keyword matching, built on Docling, OpenSearch, and Gemini.
RAG · Agents
Smaller builds and experiments, each made to learn or solve something specific. JEEnius and How Models Think grew into full offerings of their own; you'll find them in the catalog.
An agentic, RAG-based resume screener that reads resumes the way a human would instead of doing brittle keyword matching, built on Docling, OpenSearch, and Gemini.
An AI-powered financial intelligence pipeline: downloads SEC 10-K filings, parses them with Docling, runs FinBERT sentiment analysis, and produces ensemble-scored, backtested trading signals.
A conversational AI assistant for pharmaceutical sales reps: answers drug FAQs, summarizes clinical trial data, and personalizes sales pitches, built on LangChain, GPT-4, RAG, and FAISS.
A movie recommendation engine combining content-based filtering with Word2Vec embeddings and RNNs, served through a Streamlit app.
A web-based financial strategy game for 3 to 10 friends, playable start to finish in under an hour: build a company, grow a Power score across cash, real estate, stock, and rivals, and bluff your way through hostile takeovers.
A Databricks certification prep app.
Classic ML algorithms implemented from scratch rather than imported from a library, to build real intuition for how they work. The same philosophy behind How Models Think.