Portfolio Details
Each project in this portfolio reflects my hands-on experience in solving real-world problems using data science, machine learning, and full-stack development. From predictive models to intelligent web applications, these solutions are built with a focus on scalability, functionality, and impact.
.png)
.png)
.png)
Project Information
- Project Name: Perfume Haven
- Category: Recommendation System
- Model Type: Cosine Similarity (Content-Based)
- Github Repo: Link
Technologies Used
The project is powered by Python and FastAPI on the backend, with a clean and responsive frontend interface. It uses TF-IDF vectorization and cosine similarity for recommending perfumes based on scent, brand, and category. Additional tools include MLflow for experiment logging, DVC for versioning, and Docker for model serving.
Details of Project
Perfume Haven is a real-time, content-based perfume recommendation web application. It allows users to receive smart, highly relevant perfume suggestions as they type into the search bar. By analyzing a dataset of 10,000+ perfumes, the system suggests options based on scent profile, brand, and fragrance category using TF-IDF and cosine similarity.
The project showcases the integration of data preprocessing, vector-based modeling, and real-time search handling — delivering personalized results in under 300ms. It also features automated pipelines for consistent retraining and reproducibility, ensuring that the recommendation quality remains reliable.
Project Features
- Smart search bar for real-time perfume suggestions.
- Cosine similarity–based model using TF-IDF feature vectors.
- Recommendations filtered by brand, scent family, and category.
- Clean, mobile-friendly user interface integrated with FastAPI backend.
- High performance: Delivers results in less than 300ms on average.