Sapper

As a Senior Data Scientist at Sapper.ai, you will play a pivotal role in driving data-driven solutions within our Application Team. Leveraging your expertise in AI and ML, you will work on developing and implementing advanced models and algorithms to extract valuable insights from data, enhance OCR capabilities, and revolutionize processes in the Insurance and Accounts Receivable sectors. You'll be working on AI platforms, which are a set of products responsible for building, training, and deploying ML models for extraction and interpretation of semi and unstructured documents using NLP and Computer vision. Additionally, you'll be involved in ML model development and building automated training pipelines that deliver high-performance models on a given data set. Collaborating with application teams, you will understand their data extraction requirements that can be solved using ML. It will also be your responsibility to educate Subject Matter Experts on ensuring high-quality data annotations and assist them in validating your ML models. Furthermore, you will participate in building APIs around your model for production and ensure that they can deliver expected accuracies and throughput requirements. To excel in this role, you should have a strong theoretical and practical knowledge of ML model development, hyper-parameter tuning, and production deployment. You must possess significant experience in building models using libraries like Tensorflow and Pytorch. Proficiency in writing code in Python is essential, along with an understanding of well-known architecture/algorithms in NLP such as Transformers, LSTMs, and GRUs. Experience in fine-tuning pre-trained models like BERT and ELECTRA for downstream tasks like NER, Classification, etc., is also required. Additionally, familiarity with Object detection using libraries like Yolo or TF object detection API and the ability to comprehend model performance/shortcomings using various metrics like F1-score, mAP, confusion matrix, etc., are crucial. Knowledge of standard packaging and deployment solutions like Tensorflow Model Server, MLflow, or ONNX, as well as practical knowledge of libraries like Numpy, Pandas, and Spacy, will be beneficial. You should also have practical knowledge of building RESTful APIs around your model using Fast/Flask API and a strong Understanding of MLOPs - Life cycle of Machine Learning Models. It would be advantageous to have knowledge of the tf.data API for building reusable training pipelines and an understanding of various model pruning techniques to maximize production throughput and minimize resource requirements needed for deployment. This job opportunity was posted by Mobeena Syed from Sapper.,