Chris Fregly

spark spark streaming spark sql apache spark kafka big data approximations graphx mllib spark cassandra connector cassandra bloom filter recommendations matrix factorization collaborative filtering alternating least squares count min sketch hyperloglog sketch algorithms tensorflow machine learning artificial intelligence spark ml spark mllib advanced spark and tensorflow meetup gpu kubernetes tensorflow serving tensorflow distributed machine learning pipeline aws sagemaker aws keras distributed tensorflow nvidia python ml pipelines xla distributed systems high performance pytorch model deployment pipeline.ai model training data science on aws amazon web services cuda tensorflow ai model predictions kubeflow neural networks c++ google cloud platform ai pipeline tensorrt airflow tfx spark elasticsearch connector mlflow distributed training redis multi-armed bandit linear algebra redshift lambda architecture jit compiler open source continuous deployment istio json bert scikit-learn traffic routing performance tuning real-time processing textual analysis nlp netflix natural language processing performance mesos model optimization tensorflow xla docker queue java jupyter streaming data r model serving batch processing blinkdb high scalability graph processing google cloud ml fp16 azure ml dynamodb data frames orc tensorflow extended s3 xgboost quantum computing catalyst optimizer rest pandas parquet avro redis ai redis streams data sources api jupyterhub model explainability predicate pushdown ai platform numpy open data science conference model runtime tpu canary deployment aot compiler reinforcement learning hyper-parameter tuning tensorfllow cluster scheduler google cudnn cudamalloc inference quantization convolutional neural networks pascal half precision malloc k80 p100 timsort compression asynchronous cpu cache locality hash sort task scheduling saturate network partitioning throughput shuffle scala algebird best practices tuning data ingestion sampling etl tachyon monitoring kinesis big data kinesis spark streaming approximations la spark streaming kinesis aws int8 algorithms momentum gradient descent zeppelin nifi notebooks elasticsearch kafka streams ipython jdbc analytics partition pruning erlich nosql csv cache alignment merge sort optimization tungsten spill to disk sort-merge join slack zoom advanced kubeflow meetup grateful dead peter drucker flywheel oreilly media oreilly book workshop amazon kendra chatbot amazon lex text classification amazon comprehend a2i human-in-the-loop augmented ai aws deequ aws data wrangler matplotlib emr local zones outposts ec2 recap reinvent hyper-parameter optimization katib tpu kubernetes spark spark ml swift redis.ai scale ray software aws cloud feature engineering sagemaker data wrangler fairness bias model debugger model drift training-serving skew model monitoring bell pair amplitude amplification phase manipulation relative phase magnitude amplitude photons circle notation entanglement superposition quantum processing unit qpu post-quantum cryptography quantum advantage quantum supremecy quantum simulations qubit amazon braket pipeline ingestion accelerators odsc hdfs hybrid-cloud metrics high availability intelligent routing intelligent infrastructure predicting scoring real time dataset tensorboard parameter server hadoop scipy numba on-premise hybrid cloud cloud azure visualizations embedded devices compiler functional jackson streaming hive combinators tesla netty nir eyal tensorflow federated tensorflow encrypted udf privacy model tracking continuous experiments data pipeline model convergence reduced precision fp32 profiling half-precision tf.data model inference predictive analytics hardware-specific model optimization ai for ai cntk onnx mxnet chainer tenesorflow system engineering kubernetes meetup mapr object detection video classification model management a/b testing big data spain