yield · AI Memory Assistant

Full-stack second brain that unifies chat, memories, file-based RAG, and smart reminders. The frontend (Next.js 14 + Tailwind + Redux) talks to the FastAPI backend which streams Grok-3 responses, manages user-scoped Supabase data, and runs background schedulers.

---

✨ Product Highlights

• Auth-guarded dashboard with persistent chat history, memory bank, smart inbox, file uploads, and profile settings.
• Streaming AI assistant that automatically searches or mutates memories through LangChain tools.
• Memory Processor daemon that sanitizes/extracts long-term facts before writing to Supabase.
• File uploads ➜ RAG using Gemini embeddings, RecursiveCharacterTextSplitter, and Supabase Storage/vector search.
• Smart Inbox reminders powered by APScheduler + notifications API.
• Responsive landing page at / showcasing the product with framer-motion micro-interactions.

---

🧱 Tech Stack

Layer	Technologies
Frontend	Next.js 14 (App Router), React 19, Tailwind, framer-motion, Redux Toolkit
Backend	FastAPI, LangChain, xAI Grok-3, Supabase client, APScheduler
AI / Memory	Google Generative AI embeddings, Supabase vector store, custom memory tools
Storage	Supabase Postgres + Storage buckets

---

📂 Repository Layout

.
├─ backend/        # FastAPI service
│  ├─ src/
│  └─ tests/
└─ frontend/       # Next.js application
   ├─ app/
   ├─ components/
   └─ lib/

Each subdirectory ships with its own README plus package manager (pip / npm).

---

🚀 Getting Started

Backend

cd backend
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
# create backend/.env and add your Supabase/xAI/Google keys
uvicorn main:app --reload --port 8000

Frontend

cd frontend
npm install
# create frontend/.env.local with NEXT_PUBLIC_SUPABASE_* vars
npm run dev

Visit:

• Landing page: http://localhost:3000/
• Dashboard (auth required): http://localhost:3000/chat
• API docs: http://localhost:8000/docs

---

🧪 Testing & Quality

• Backend smoke tests live under backend/tests (pytest).
• Manual verification: streaming chat, memory add/delete, file ingestion notifications, smart inbox badge, responsive layout.
• Structured logging: LOG_LEVEL controls the new logging pipeline; every critical service logs via yield.* namespaces.
• CI-ready: add your preferred GitHub Action calling pytest + npm run lint to ship this to recruiters/teams.

---

🛠 Helpful Scripts

Command	Description
uvicorn main:app --reload	Backend dev server
npm run dev	Frontend dev server
pytest	Backend smoke tests
npm run lint	Next.js linting (if configured)

---

🔍 Observability

• Central logging config (backend/src/core/logging.py) emits consistent console output.
• Scheduler, file ingestion, memory processor, and AI service all log structured events.
• Health endpoint (/health) is test-covered and safe to expose for uptime monitors.

---

📌 Portfolio Notes

What’s production-ready:

• Full auth pipeline (Supabase JWT) and user-scoped data access.
• Background sanitization and proactive memory retrieval for better AI answers.
• File-based RAG integration with chunked embeddings.
• Modern landing page plus responsive dashboard pages.

What to mention as next steps:

• Broader automated test coverage (service-layer and frontend component tests).
• Rate limiting & telemetry/export to a log drain.
• Additional CI automation (lint + type checks).

This repo now tells a complete story for advanced product engineering work—feel free to reference it directly in your portfolio or attach screenshots/video walkthroughs for extra context.

---

Built by Jesse’s “yield” initiative • 2025