Edric Gan

Projects

Selected products and systems.

I work across product, backend, frontend, AI, and delivery. These projects show the range.

AI Tech Advisor

AI platform for social media

One workflow to plan, create, review, and publish social content.

  • 4major social platforms
  • 4+image providers
  • Image + videogeneration pipelines

I advise the team on AI architecture, agent workflows, product integrations, and delivery.

  • Agent orchestration: LangGraph manages multi-step content generation, tool use, and review flows.
  • Workflow automation: n8n connects services and runs repeat jobs outside the main application.
  • Media routing: Higgsfield, fal, Grok, Nano Banana, and other providers support image work. A separate pipeline handles video generation.
  • Publishing: The product connects with Facebook, Instagram, X, and LinkedIn.
  • Marketing workflow: Brand context, drafts, approvals, scheduling, and analytics live in one place.

This architecture reduces manual marketing work. It also lets the team compare models, control costs, and add providers without rebuilding the product.

LANGGRAPH · N8N · NEXT.JS · VERCEL · HIGGSFIELD · FAL · GROK · NANO BANANA · VIDEO GENERATION

Current · Live

Lingooso

Turn any YouTube video into practice you can use every day.

  • 60+early users
  • 10languages
  • AIpractice and shorts

Lingooso creates vocabulary, flashcards, quizzes, synced reading, and short practice videos from real content. AI explains each word in context and generates useful exercises.

Engaging, bite-sized shorts help learners practice often without a long study session. Spaced repetition brings each word back at the right time.

More than 60 early users are testing the product and shaping what comes next.

Visit Lingooso
AI PRACTICE · LEARNING SHORTS · VIDEO PROCESSING · SM-2 · TEXT TO SPEECH · ANALYTICS
Lingooso language lesson product preview
Language lessons built from YouTube content.

Open source

TEAMMATES

A peer feedback platform used by universities.

I contributed to a mature open-source product with a large user base and an established engineering process.

The work strengthened my skills in shared code ownership, review, and safe changes.

View the repository
TEAMMATES open-source project
Peer feedback for teaching teams and students.

Data science

Prediction and spam detection

Three projects that move from data checks to measured model results.

  • 15%over baseline
  • 3stacked models
  • 2spam approaches

The flat-price model uses MLflow and a stack of random forest, Elastic Net, and XGBoost. It beat the baseline by 15%.

The spam work compares classic machine learning with GloVe embeddings and a bidirectional GRU.

Read the case study
Singapore flat price prediction interface
Experiment tracking and model stacking for flat prices.