TensorFlow Lite Model Conversion from Robo Flow with existing datasets

I'm on the lookout for a skilled ML developer who can assist me in converting my existing pre-trained model, which is currently being used with Roboflow, into a TensorFlow Lite model. This conversion is to enable the model to perform inference in offline mode on images captured from smartphone cameras. The goal is to have the model running at near real-time speeds on both Android and iOS smartphones. The dataset has 12 classes of objects of numeric digits only.

Key Requirements:
- Convert a pre-trained model from Roboflow to TensorFlow Lite
- Implement the model for offline inference.
- Achieve near real-time speed on both Android and iOS
- The captured images to be uploaded to cloud when internet connection is available.
- MUST have TensorFlow Lite experience with evidence required.

Ideal Skills and Experience:
- Proficiency in TensorFlow Lite and experience with model conversion
- Previous experience with deploying models on both Android and iOS
- Understanding of image processing and optimization for real-time inference in offline smartphones.

Do NOT bid for this project if you're unable to provide example of previous work specific to TensorFlow Lite. It must be included in bid. Otherwise, bid will automatically be disqualified