Datan lisäys Jobs

I have a large collection of raw audio recordings and need an expert who can turn them into a high-performing voice model inside Amazon SageMaker. Your job starts with designing an efficient preprocessing pipeline—cleaning, segmenting, and augmenting the audio so that the data is ready for distributed training on SageMaker. Once the data is prepared, I’d like you to select or build a state-of-the-art architecture, train it end-to-end, and fine-tune until we hit 90-95 % word-level accuracy on my validation set. Please incorporate best-practice techniques such as mixed-precision training, hyper-parameter tuning jobs, and automatic model versioning so we can reproduce results later. Finally, the trained model must be packaged as a SageMaker endpoint that plugs directly into ...

DCP Global is scaling up a large-volume language data project and needs reliable help with three core activities: tagging, grading, and annotating content. You’ll be working inside our client's proprietary browser-based tool. Accuracy and consistency are paramount. Scope • Weekly commitment: at least 20 hours until completion. • Pay rate: USD $13 per hour paid in a monthly cycle for all accumulated tasks. Deliverables – Correctly tagged linguistic units following the project’s taxonomy. – Graded data sets with clear pass/fail or scaled quality scores. – Detailed annotation notes where required, using the tool’s inline comment feature. Acceptance criteria 1. 97 % or higher agreement with gold-standard samples during spot checks. 2. All w...