stack.config.json

const website = {

backend: [

"FastAPI",

"LlamaIndex",

"Qdrant",

"Groq"

],

frontend: [

"Next.js 16",

"React 19",

"TypeScript"

],

infra: [

"Docker",

"Render",

"GitHub Actions"

]

}

Ask about my career

I'll answer based on my real experience and projects.

$marino-rag --interactive

▶ready

› ask me_

✓ Recent Updates

v1.4 Upgraded embedding to bge-base-en-v1.5 (768 dim) via HuggingFace remote API — better semantic quality, zero RAM on server
v1.3 Switched LLM to qwen/qwen3.8-27b on Groq — date-aware prompt for accurate age/duration answers
v1.2 Conversation memory with rolling window
v1.1 Docker image containerized and deployed on Render
v1.0 Qdrant vector storage moved to local (no cloud cluster costs)

→ Planned Improvements

GitHub Actions CI pipeline — build Docker image with vector storage generated on-the-fly (never committed to repo)
Better UX for slow cold starts (server warm-up feedback)

</> Source Code

Backend built with FastAPI · LlamaIndex · Qdrant · Groq

View the source code here →