01

PeakTranslate Live

AI-powered video and document translation platform — peaktranslate.com

PeakTranslate is a subscription-based SaaS that lets users upload videos or documents and receive translated subtitles, AI-generated voice-over dubbing, and translated documents. It handles the full pipeline from media ingestion to delivery.

Architecture

Client (React + Vite) │ ├──► FastAPI Server ──► Celery Workers ──► Result │ │ │ │ ├── Auth (JWT) ├── DeepL (Translation) │ ├── Stripe Billing ├── OpenAI Whisper (Transcription) │ └── PostgreSQL ├── ElevenLabs (Voice Dubbing) │ └── DO Spaces (Storage) │ └──► Redis (Queue + Cache)

Tech Stack

React TypeScript Vite shadcn/ui FastAPI Celery Redis PostgreSQL Alembic Stripe DeepL ElevenLabs DO Spaces

Key Technical Decisions

Backend Highlights

Database Schema

Core models: User, Job, UserSubscription, SubscriptionPlan, BillingEvent, CreditTransaction. Normalized with proper relationships and constraints.

Deployment

Frontend
Vercel / Lovable
API
api.peaktranslate.com
Database
DO Managed PostgreSQL
Storage
DO Spaces (Frankfurt)

02

PHOENIX Quant AI Live

Stock signal intelligence terminal — stocksphoenix.tech

An automated system that monitors Twitter/X and Reddit for stock ticker mentions, scores sentiment, generates trading signals, manages a simulated $100K portfolio, and produces AI-written equity analyst reports — all served through a real-time dashboard.

Architecture

Scheduler (every 60s) │ ├── Collectors ──► Twitter (GraphQL scraping) │ Reddit (public JSON API) │ ├── Processors ──► Aggregator (mention scoring) │ Signal Engine (BUY/SELL logic) │ Filter (market cap, volume, news) │ Paper Trader (TP/SL/time exits) │ ├── Scanners ───► Yahoo Finance losers │ Risk filter (L-shape, neg. news) │ ├── Synthesizer ─► OpenRouter LLM analyst reports │ └── Notifiers ──► Telegram alerts FastAPI API Server ◄──► Next.js Dashboard │ PostgreSQL (10+ tables, 50K+ mentions)

Tech Stack

Next.js 16 React 19 Tailwind CSS Lightweight Charts FastAPI Python PostgreSQL VADER Sentiment OpenRouter yfinance Telegram API

Signal Generation Pipeline

  1. Collection — Scrapes configured Twitter accounts and Reddit subreddits every minute. Extracts $TICKER mentions with deduplication.
  2. Filtering — Validates tickers against a 400+ term blocklist, then checks market cap (>$500M), price (>$2), and volume (>500K) via yfinance.
  3. Scoring — Aggregates mention frequency, source diversity, and sentiment into a composite score using logarithmic weighting.
  4. Signal — BUY when score exceeds threshold with sufficient source diversity. SELL when sentiment turns negative on held positions.
  5. Paper Trading — Enters positions with 2% of virtual capital. Exits on 10% take-profit, 5% stop-loss, or 5-day time limit.

Dashboard Features

AI Reports

On-demand equity analysis via OpenRouter LLM. The synthesizer concurrently fetches financials, news, insider trades, and relative performance, then builds a structured prompt for the LLM to produce a full analyst report covering business profile, catalysts, risks, and capital decision.

Deployment

Server
VPS (Nginx + SSL)
Process Manager
PM2 (3 services)
Database
PostgreSQL (50K+ rows)
SSL
Let's Encrypt (auto-renew)

03

Nihongo.page Live

Japanese vocabulary learning platform — nihongo.page

A personalized Japanese language learning app built around spaced repetition. Users study 10 vocabulary words per day through interactive flashcards, get AI feedback on practice sentences, and listen to weekly generated audio scripts tailored to their vocabulary level.

Architecture

Next.js 16 (App Router — SSR + API Routes) │ ├── Auth ──────► NextAuth v5 + Google OAuth ├── Database ──► Prisma ORM ──► PostgreSQL ├── AI ────────► OpenRouter (Qwen3) — sentence review ├── TTS ───────► Alibaba DashScope (Qwen3-TTS) └── Storage ──► DO Spaces (weekly audio files)

Tech Stack

Next.js 16 React 19 TypeScript Prisma PostgreSQL NextAuth v5 OpenRouter Qwen3-TTS Tailwind CSS DO Spaces Umami Analytics

Spaced Repetition System

Custom SM-2 variant with three learning phases:

The algorithm tracks interval, easeFactor, repetitions, dueDate, and nextShowAt per user-card pair. Cards rated "Again" or "Hard" are re-inserted into the current session queue immediately.

Weekly Audio Scripts

Each week, the system generates 3 audio scripts using the user's learned vocabulary:

Users choose script types: Story, Dialogue, News Broadcast, Travel Scene, or Action Adventure. Text is chunked by Japanese punctuation (max 300 chars), sent to Qwen3-TTS in parallel, concatenated, and stored on DO Spaces.

Database Schema

10 Prisma models: User, Card (2,000+ entries from Kiashi deck), UserCard (SRS state per user), DailySession, UserSentence (with AI feedback), WeeklyScript (with audio URLs). Part-of-speech enum: Noun, Verb, Adjective, Adverb, Particle, Other.

Features