A collection of AI-powered tools and full-stack systems — LLM agents, RAG pipelines, smart resume scoring, and automation built with real intent.
An ATS-scoring resume builder that grades a resume against a job description with a heuristic 0–100 score across contact info, section coverage, action verbs, quantified results, keyword match, and format. Surfaces missing keywords and supports LLM-assisted summary generation, with multiple export-ready templates.
A set of autonomous AI agents built on OpenAI function calling — each agent has a tool belt, memory, and a task planner. Includes a RAG pipeline for document Q&A using vector embeddings and a retriever chain for long-context reasoning.
Selenium-driven web scrapers and Google Maps API extractors for bulk business data collection. Handles dynamic content, pagination, rate limits, and exports clean CSVs — built to run headless in the background.
Beyond shipping the projects above, I'm actively working through the AI engineering stack — the layer that turns a model into a product. These are the topics I'm learning and applying right now. The full list lives on the hub → Practice.
Tokenisation, context windows, temperature & sampling, system vs user prompts, structured outputs, JSON mode, function calling.
in progress · buildingEmbeddings, vector stores (FAISS, Chroma, pgvector), chunking strategies, re-ranking, retrieval evaluation, hybrid search.
building · rag-assistantTool use, function-calling loops, ReAct pattern, multi-agent orchestration, planning, memory, the agent–environment loop.
studying · nextChain-of-thought, few-shot, role prompting, output schemas, evaluation harnesses, prompt versioning and regression testing.
applied · dailyCosine similarity, semantic search, ANN indexes, hybrid lexical + vector retrieval, metadata filtering, embedding-model selection.
exploringHallucination detection, guardrails, observability, LangSmith-style tracing, evals, red-teaming, prompt-injection mitigation.
queued