TensorFlow for FinTech - NYC MDEC Sprint

Slide 1

Slide 1 text

TensorFlow for FinTech And scaling your applications on Google Cloud Vikram Tiwari - GDE, Google Cloud Platform

Slide 2

Slide 2 text

Intelligence Everywhere

Slide 3

Slide 3 text

Analysis Portfolio Management Sentiment Analysis Customer Segmentation Live Support/Chat Bots Prediction Fraud Detection Credit Scoring Customer Retention Search Engine

Slide 4

Slide 4 text

Usage in FinTech

Slide 5

Slide 5 text

Mining Forms, Documents etc Scanned Images Audio Calls Video Data

Slide 6

Slide 6 text

Stocks Risk Credit scores Products Classification

Slide 7

Slide 7 text

Security Malicious behaviour Fraud prevention Data Security Compliance

Slide 8

Slide 8 text

- Engaged and Evolving Community - Rapid Development - Distributed Training - Tensorflow Serving - Prebuilt models, TPUs Why Tensorflow?

Slide 9

Slide 9 text

Easy to hire Code maintainability Supporting community Prebuilt models More research, more gains Community

Slide 10

Slide 10 text

tf.contrib.learn Keras Scikit-Learn style API API Documentation TensorBoard Development

Slide 11

Slide 11 text

Graph Execution Model Write once, train anywhere Train on clusters or individual CPUs Extensive GPUs support Training

Slide 12

Slide 12 text

Continuous Training Pipeline Versioned Models Easily use resulting models gRPC or REST based APIs Serving

Slide 13

Slide 13 text

Stable Version Tensorflow Processing Units (TPUs) Custom Filesystems Tensorflow Dev Summit AutoML More...

Slide 14

Slide 14 text

No content

Slide 15

Slide 15 text

- Leader in Data and Machine Learning innovation - Ready-to-use Machine Learning APIs - High performance and cost effective products and services - Tensorflow on TPUs for best performance - Regulatory and Compliance ready Why Google Cloud?

Slide 16

Slide 16 text

Fastest growing cloud platform with a major focus on Open Source

Slide 17

Slide 17 text

Storage Persistent Disk Datastore, Cloud SQL, BigTable Firebase Cloud Spanner, BigQuery Different tiers for Cloud Storage Services Compute Engine (GCE) App Engine (GAE) Kubernetes Engine (GKE) Cloud Functions (GCF) Load Balancing, IOT core, Dataflow

Slide 18

Slide 18 text

Wide range of compute platforms to choose from

Slide 19

Slide 19 text

Thank you! Learn more: - Tensorflow - Tensorflow resources - Google Cloud Machine Learning APIs - Google Cloud Platform Solutions @Vikram_Tiwari - https://vikramtiwari.com