Python Code Hierarchy Model Integration

Urgent Freelancer Job: AI Engineer for Hierarchical Classification & _Val Normalization ? Job Title: ? AI Engineer Needed ASAP: Implement Hierarchical Classification with _Val Normalization & OCR Learning ? Job Description: We need an AI/ML Engineer immediately to integrate a hierarchical text classification model that: ✅ Learns hierarchical label structures without regex or hardcoded rules. ✅ Handles _Val normalization by predicting numerical values alongside categorical labels. ✅ Learns OCR correction dynamically from the input text (no manual regex cleaning). ✅ Trains on full hierarchical context to understand document structure and reference relationships. ✅ Processes multiple surveys within a single Full_Text input, maintaining context. ? Timeline: ASAP – We need someone who can start immediately! ? Message me with your cost and timeline after reviewing the current code! ? Project Scope: ✅ Tie into our existing hierarchical classification system (see GitHub below). ✅ Ensure _Val labels are correctly predicted and stored. ✅ Model must learn to ignore OCR errors, not rely on regex-based cleaning. ✅ Batch sizes must be structured around Full_Text entries and their children (ensuring sequential order). ✅ Multi-label classification must maintain parent-child dependencies. ✅ Deploy the trained model for inference on raw .txt inputs. ? Current Work Done & Repository Details We have a fully structured hierarchical classification system, available at: ? GitHub Repository: Hierarchy Trainer ? What’s Already Implemented? ✔ Structured hierarchical label extraction from raw .txt files ✔ Bidirectional sibling relationships within the hierarchy ✔ Graph-based hierarchical representation of classification results ✔ GUI for training, testing, and visualization ✔ CUDA support for GPU acceleration ✔ Standardized _Val labels tracking numerical values ? What Needs to Be Done? ? Extend the current system to support _Val normalization (predict numerical values as well as labels). ? Ensure batch sizes are structured around Full_Text labels and their children (no random shuffling). ? OCR errors must be learned dynamically by the model, not corrected with regex. ? Train the model with full hierarchical context to ensure document structure is learned. ? Ensure multiple surveys within a single Full_Text are recognized and classified correctly. ? Optimize the pipeline for structured multi-label predictions. ? Deploy the trained model to work seamlessly within our current framework. ? Important Details About the Data & Training Process ? Each Full_Text label is a real life example of the expected raw .txt input for inference, but there will be variations in the noise and unimportant text in real life scenarios, so it is imperative the model only extracts the data it needs. ? The model must be trained to process multiple surveys within a single Full_Text block. ? It must learn the structure of survey documents, not just individual labels. ? Hierarchical relationships are already established in the [login to view URL] and GUI—your task is to transform them into a model-compatible format. ? Batching is already set up in the [login to view URL] to group Full_Text labels with their children in order—please confirm it aligns with the training process. ? Ideal Skills & Experience: ? Experience with Hugging Face models (DeBERTa, BERT, or hierarchical transformers). ? Deep learning for text classification (PyTorch, TensorFlow). ? Handling hierarchical multi-label classification. ? Working with numerical normalization & structured label dependencies. ? Understanding how to train models that learn OCR correction dynamically. ? Experience integrating models into existing AI pipelines. ? Budget & Timeline ? Competitive, based on experience ⏳ Project Timeline: ASAP – Send me your cost and estimated time to complete after reviewing the code. ? How to Apply? ? Send a message with: A brief summary of your experience with hierarchical classification & numerical normalization. Examples of similar projects you've worked on. Your cost estimate & timeline after reviewing the GitHub repository. We need someone who can move fast. If you're confident in your ability to deliver, message me ASAP! ? GitHub Repository: [login to view URL] Looking forward to working with you! ?