Video Content Recommendation Engine
Budget: ₹12,500 – ₹37,500 INR
I want to build a content-based recommendation engine that serves up the right videos to the right user at the right moment. The core need is an end-to-end system—data ingestion, model training, and an API (or micro-service) that returns ranked video suggestions in real time.
My dataset will include video metadata, user interaction logs, and basic demographic tags; you are free to suggest additional signals if they will improve accuracy. I’m open to classical approaches (collaborative filtering, matrix factorization) as well as deep-learning architectures such as two-tower models or sequence-aware networks. What matters is measurable lift in click-through and watch-time.
Please send a detailed project proposal that covers:
• Your chosen algorithms or model stack and why
• A high-level data schema or feature plan
• Milestones from proof of concept to production deployment
• How you will benchmark success (accuracy metrics, A/B plan, etc.)
If you have past work with recommender systems, feel free to reference it, but the proposal itself is the deciding factor. A concise timeline and the tools you expect to use—Python, TensorFlow/PyTorch, Spark, or other—will help me evaluate fit. I will be ready to supply sample data once we agree on the approach, and I’m aiming for a first working model within a few weeks of kickoff.
My dataset will include video metadata, user interaction logs, and basic demographic tags; you are free to suggest additional signals if they will improve accuracy. I’m open to classical approaches (collaborative filtering, matrix factorization) as well as deep-learning architectures such as two-tower models or sequence-aware networks. What matters is measurable lift in click-through and watch-time.
Please send a detailed project proposal that covers:
• Your chosen algorithms or model stack and why
• A high-level data schema or feature plan
• Milestones from proof of concept to production deployment
• How you will benchmark success (accuracy metrics, A/B plan, etc.)
If you have past work with recommender systems, feel free to reference it, but the proposal itself is the deciding factor. A concise timeline and the tools you expect to use—Python, TensorFlow/PyTorch, Spark, or other—will help me evaluate fit. I will be ready to supply sample data once we agree on the approach, and I’m aiming for a first working model within a few weeks of kickoff.