Lead Data Scientist
Current- Created automated big data pipelines using apache hive for data science, analytics, and big data projects.- optimizing data storage, retrieval, and querying.- preparing data for ingestion in reporting software’s like tableau- creation of dashboards using tableau- conducted research on botnet data for identifying malicious DDoS attack activity.- Assisted in requirements gathering and translating them to production code.- Optimized and maintained existing machine learning models for fingerprint image classifications, such as CNNusing Python’s Keras and tensorflow to detect falsified fingerprints- Created TFRecord image data ingestion pipeline using tf.Data API, making them scalable and enabling training ofmodels with datasets larger than CPU memory- Automated training and optimization of model hyperparameters to quickly conduct and test 50 different variationsof the model, storing the results and generating automated reports- Created tooling and scripts to further data exploration efforts for other data scientists, helping them become moreeffective at exploring the data and other tasks- Productized existing TensorFlow models by converting them to tflite format, allowing integration with existing C++and android applications- Conducted Transfer Learning using ResNet 50 pretrained model by freezing the bottom layer and retraining toplayers with fingerprint images- Achieved Half Precision (FP16) training on Nvidia Titan RTX and Titan V GPU for TensorFlow 1.14