Contact person: Ingredient Substitution ML Module

Phone:Show

Email:Show

Location: Meridian, United States

Budget: Recommended by industry experts

Time to start: As soon as possible

Project description:
"I’m enhancing my mobile recipe app with a smart ingredient-substitution feature and need a machine-learning specialist to build the core module for it. The goal is simple: when a user is missing an item—say, buttermilk or shiitake mushrooms—the app should instantly propose realistic alternatives that keep flavour, texture, and nutritional value as close as possible to the original recipe.

Here’s how I picture the workflow. You’ll source or curate an appropriate dataset (public, licensed, or synthetically expanded), train and fine-tune a model that captures culinary pairings and constraints, then expose it through a lightweight API or an on-device library that my app can call. Response time has to feel real-time for the user, and the model must be able to improve as new usage data arrives. Any mainstream ML stack is fine—TensorFlow, PyTorch, or even a managed service—so long as deployment fits within typical mobile limits.

In your reply, focus on your practical experience building and shipping similar recommendation or substitution engines, especially where food data or flavour profiles were involved. If you’ve solved cold-start problems, handled limited edge-device resources, or balanced model accuracy with speed, that’s exactly the kind of insight I’m after.

Deliverables will be:
• Trained model files and training code
• Inference endpoint or embeddable library with clear integration instructions
• Brief documentation describing data sources, evaluation metrics, and retraining workflow

If you can get users cooking happily with whatever is already in their pantry, we’re a match." (client-provided description)