Project← selected work
RecipeSense: ML based Recipe Recommender
Recommendation engine built around a custom Hybrid-Weighted similarity kernel that fuses TF-IDF text embeddings with structured metadata signals — outperforms off-the-shelf collaborative filtering in cold-start scenarios. Indexes 7,000+ recipes and uses T-SNE to visualize cluster separation between cuisine types.
Overview
This Recipe Recommendation System uses advanced machine learning techniques to suggest personalized recipes. It employs TF-IDF vectorization to understand user preferences and provides tailored cooking suggestions through an intuitive interface.
Key Features
- Personalized recipe recommendations
- Ingredient-based search
- Dietary restrictions filter
- Cooking time estimation
- Nutritional information
Stack
frontend
React.jsMaterial-UI
backend
PythonFlaskscikit-learn
database
PostgreSQL
ml
TF-IDFCosine Similarityscikit-learn
Challenges & Solutions
- Creating accurate recommendation algorithms
- Handling diverse recipe data formats
- Optimizing search performance
Gallery