Deep dive
Read the full case study
The technical details, the tradeoffs, and the results — written down properly.
Read case study Back to projects
AI & Machine LearningFinanceongoing
VietProStocks
Local-first stock research for the Vietnamese market: FAISS retrieval with validated citations, technical indicators and Telegram alerts — engineered to run on a Raspberry Pi.
2025 4 months 1 team members
top-K 6–12 → 3–5
retrieval
Raspberry Pi 4 8GB
hardware
local Q4 7B
model
URL-validated
citations
Overview
VietProStocks combines Vietnamese market data, technical analysis and a retrieval-augmented LLM into a research assistant. The RAG stack — MiniLM embeddings, a FAISS index and a local llama.cpp model — is designed to run entirely on a Raspberry Pi 4 8GB, with an OpenAI-compatible backend as an escape hatch when more compute is available.
The challenge
LLM research tools are expensive and hallucination-prone. This one had to produce cited answers on a $100 board, respect free-tier budgets, and never present an invented source as evidence.
Our solution
Chunked a curated news and document corpus (500 tokens / 100 overlap), embedded it with all-MiniLM-L6-v2, indexed with FAISS, and reranked hybridly — semantic similarity, recency and source trust — down to three to five cited chunks. A multi-backend LLM client, per-user call caps and URL-validated citations keep answers bounded and honest.
Key features
Retrieval-augmented Q&A with citations and confidence scores
Technical indicators: SMA, EMA, RSI, MACD, Bollinger Bands
Vietnamese market price and news ingestion
Smart alerts with unread notification counts
Trilingual interface (VI, EN, JA)
Raspberry Pi deployment with cron backups
Results & impact
Cited answers served from a Raspberry Pi 4 8GB
Citations validated against source URLs before display
Hybrid rerank keeps fresher, more trusted sources on top
Scheduled alerts through a Telegram bot and Slack webhooks
Zero-cost operations: systemd + Cloudflare Tunnel + daily backups
Multi-backend LLM client: llama.cpp, OpenAI-compatible, mock
Related Projects
Explore more projects similar to this one
AI & Machine Learning
2025
SentinelVision Edge
SentinelVision Edge turns low-cost Raspberry Pi cameras into an AI security fleet. Node agents run YOLO person and face detection at the edge, publish occupancy and dwell-time signals over MQTT, and stream video through MediaMTX. A FastAPI backend with PostgreSQL, Redis and Celery manages cameras, areas, recordings and roles, with a Vue 3 dashboard for real-time status.
Raspberry PiYOLOUltralytics+10
View Project
Document Intelligence
2025
DocuVerse
DocuVerse ingests PDFs and images, runs DeepSeek-OCR through a vLLM-served model, and delivers Markdown-first documents inside a workspace model with role-based access. A FastAPI/PostgreSQL/Redis/Celery backend handles OCR jobs with live WebSocket progress, and the same API powers both a Vue web client and a Tauri desktop build.
PythonFastAPISQLAlchemy+8
View Project
AI & Machine Learning
2026
NihonGO! AI Japanese Tutor
NihonGO! pairs learners with 3D avatar characters for one-on-one Japanese conversation practice. A streaming tutor replies in pure Japanese while a separate analysis endpoint returns corrections, vocabulary and grammar notes. The product spans a Flutter app (mobile, web, desktop), a Hono/TypeScript backend on Supabase, and four LLM providers behind one interface — with an internal security audit completed before store submission.
FlutterDartRiverpod+9
View Project
Start Your Project
Tell us about your project and we'll get back to you within 24 hours.
Reference Project:
VietProStocks