SlideShare a Scribd company logo
1 of 28
1
01
Table of contents
Smava business in a nutshell
02 Why ML and AI?
03 Production environment
04 Deploying to AWS
05 Summary
2
Smava business model
01
3
4
A comparison platform for personal loans
1. 1000-50000 Euros,expanding into larger size loans (i.e. mortgage).
2. Financing for purchasing a car, repairing your home, travelling etc.
3. Provided by one of the partner banks
4. Needs to be paid back
○ Customer’s creditworthiness
○ Fundamentally different to a typical ‘buy and forget’ comparison
platform
5
Business Model
smava enables borrowers to get the best deal for consumer loans,
by pooling offers from a few dozen banks in a single website
6
Customer journey
1. www.smava.de - fill in the details online
○ An average customer has to answer 50 questions ~ 15 mins
2. Smava sends an API request to partner banks and gets a response ~2 mins
○ As soon as the registration route is finished partner banks are requested and offers
are displayed
3. Provide proof of income and other details ~ 7 days
○ Usually scans of income slips, bank statements, etc.
4. Get the money into your bank account or get a rejection.
7
8
Why ML and AI?
02
9
10
Problem statement
● Banks have varying requirements and preferences reflected by their risk
management policy
○ Minimise defaults, maximise ROI
● Banks incur costs when assessing leads from Smava
● Customer needs to qualify for the loan
○ How to efficiently match between customer and a bank?
● Customer only cares about the cost of the loan and the ease of getting it
Why machine learning?
● Uncertainty - what is the probability that customer x will get accepted by bank y?
● Ignore uncertainty
○ request all banks for each individual customer
○ Banks don’t like waste (money spent on leads that did not convert)
● Address uncertainty with heuristics
○ High income customers - Bank A, low income customers - Bank B
○ Not efficient, needs continuous updating and manual tweaking
○ Could potentially be more harmful to the business than ignoring uncertainty
● Address uncertainty with Machine Learning
○ Use the full set of data features
■ To learn and predict bank answers
○ Retrain models frequently
○ Use predictions for efficient matching (how?!)
● Blend of heuristics and Machine Learning
○ Heuristics were there first and are believed to be useful
○ Some very simple rules can be useful
11
smava Master Presentation Template - Introduction
12
Smava marketplace
● Efficient matching
○ Customer mostly care about the interest rate
○ Banks care about number of requests they receive vs the number of
payouts they provide. Banks do not like waste
○ smava’s mission is to assist the customer find the best deal on the
marketplace
■ smava is concerned both with the interest rate of the offer AND
the realisability of the offer (measured by probability to get the
offer)
Production environment
and model retraining
03
13
Smava engineering infrastructure
14
Data collection,
processing saving
(Smava website,
databases etc.)
Predictions,
portfolio creation,
execution schema
(Machine
Learning
microservice)
Bank API
connectivity
Bank_1
Bank_2
Bank_N
customer
data
request
execution
schema
offer request - response
offer request - response
~15 mins ~1 sec ~60 sec t
Design principles
● Close collaboration with engineering
○ Impact on all major KPIs
● Extensive tests
○ Unit tests
○ Statistical tests
● Automatic retraining - weekly/daily
● Live monitoring
○ input data
○ output data
15
Automatic retraining
● Get data from all sources
● Clean data and prepare it for modelling
● Train models
● Run tests
● Compute predictive accuracy on unseen data
● Return a retraining report
16
Monitoring
● Real time model predictive performance monitoring - are we predicting well?
● Anomaly detection - has anything in the outside world changed?
17
18
Tech stack
● Python for data pipelines
● R for modelling and production code
● Shiny for monitoring
● C for some internal packages
19
Predicting and using the predictions
● XGboost for predicting probabilities
● Using probabilities to guide decision making
● Intellectual property
Model hosting on AWS
04
20
21
● AWS Sagemker: platform to deploy ML model in cloud
○ Competitors: Azure AI, Google Cloud AI
● Fully managed
○ autoscaling, health checks, load balancing
○ allows to change types and number of instances; no
downtime
● Can deploy custom Docker image
● Next iteration : AWS Lambda
Model deployment
22
● Dockerized REST API (using Nginx, Gunicorn and Flask)
● SM expects server to respond to /invocations and /ping (port 8080)
○ /invocations: accept POST requests and returns prediction
○ /ping: accept GET and returns 200 (for health checks)
Model deployment
23
23
23
● Amazon cloudwatch:
○ latency, invocations, cpu-utilization,..
Monitoring
24
24
● Create a new endpoint and put in shadow for a day
● Live monitoring of shadow
Blue-Green deployment
Prod
Shadow
S3
Further
Processing
Input
output
output
25
● DS team fully owns the Application Engine microservice
○ Complete ownership of infrastructure
○ DS outputs are provided via a REST api
Application Engine microservice
Summary
05
26
27
● ML component within smava is the very core of the production system
● Opportunities to significantly impact the business in real time
○ Big potential for both improvement, but also for causing damage
● Hand in hand work with engineering, product owners and wider stakeholders in
the company
○ Advocacy
○ Explanations
○ Conflicts of interest
Summary
Thank you for your attention
Evgeny Savin
Senior data scientists
28
evgeny.savin@smava.de

More Related Content

What's hot

H2O World - Machine Learning at Comcast - Andrew Leamon & Chushi Ren
H2O World - Machine Learning at Comcast - Andrew Leamon & Chushi RenH2O World - Machine Learning at Comcast - Andrew Leamon & Chushi Ren
H2O World - Machine Learning at Comcast - Andrew Leamon & Chushi RenSri Ambati
 
Dataiku hadoop summit - semi-supervised learning with hadoop for understand...
Dataiku   hadoop summit - semi-supervised learning with hadoop for understand...Dataiku   hadoop summit - semi-supervised learning with hadoop for understand...
Dataiku hadoop summit - semi-supervised learning with hadoop for understand...Dataiku
 
Real-time Big Data Analytics: From Deployment to Production
Real-time Big Data Analytics: From Deployment to ProductionReal-time Big Data Analytics: From Deployment to Production
Real-time Big Data Analytics: From Deployment to ProductionRevolution Analytics
 
Leveraging Open Source Automated Data Science Tools
Leveraging Open Source Automated Data Science ToolsLeveraging Open Source Automated Data Science Tools
Leveraging Open Source Automated Data Science ToolsDomino Data Lab
 
Danny Bickson - Python based predictive analytics with GraphLab Create
Danny Bickson - Python based predictive analytics with GraphLab Create Danny Bickson - Python based predictive analytics with GraphLab Create
Danny Bickson - Python based predictive analytics with GraphLab Create PyData
 
Using H2O for Mobile Transaction Forecasting & Anomaly Detection - Capital One
Using H2O for Mobile Transaction Forecasting & Anomaly Detection - Capital OneUsing H2O for Mobile Transaction Forecasting & Anomaly Detection - Capital One
Using H2O for Mobile Transaction Forecasting & Anomaly Detection - Capital OneSri Ambati
 
H2O for Medicine and Intro to H2O in Python
H2O for Medicine and Intro to H2O in PythonH2O for Medicine and Intro to H2O in Python
H2O for Medicine and Intro to H2O in PythonSri Ambati
 
Introduction to Big Data: Smart Factory
Introduction to Big Data: Smart FactoryIntroduction to Big Data: Smart Factory
Introduction to Big Data: Smart FactoryJongwook Woo
 
From Machine Learning to Learning Machines: Creating an End-to-End Cognitive ...
From Machine Learning to Learning Machines: Creating an End-to-End Cognitive ...From Machine Learning to Learning Machines: Creating an End-to-End Cognitive ...
From Machine Learning to Learning Machines: Creating an End-to-End Cognitive ...Spark Summit
 
Data scientist enablement dse 400 week 8 roadmap
Data scientist enablement   dse 400   week 8 roadmap Data scientist enablement   dse 400   week 8 roadmap
Data scientist enablement dse 400 week 8 roadmap Dr. Mohan K. Bavirisetty
 
Big Data and Predictive Analysis
Big Data and Predictive AnalysisBig Data and Predictive Analysis
Big Data and Predictive AnalysisJongwook Woo
 
Cloudera Breakfast: Advanced Analytics Part II: Do More With Your Data
Cloudera Breakfast: Advanced Analytics Part II: Do More With Your DataCloudera Breakfast: Advanced Analytics Part II: Do More With Your Data
Cloudera Breakfast: Advanced Analytics Part II: Do More With Your DataCloudera, Inc.
 
Agile Data Science by Russell Jurney_ The Hive_Janruary 29 2014
Agile Data Science by Russell Jurney_ The Hive_Janruary 29 2014Agile Data Science by Russell Jurney_ The Hive_Janruary 29 2014
Agile Data Science by Russell Jurney_ The Hive_Janruary 29 2014The Hive
 
Data Science in the Real World: Making a Difference
Data Science in the Real World: Making a Difference Data Science in the Real World: Making a Difference
Data Science in the Real World: Making a Difference Srinath Perera
 
Retrieving Visually-Similar Products for Shopping Recommendations using Spark...
Retrieving Visually-Similar Products for Shopping Recommendations using Spark...Retrieving Visually-Similar Products for Shopping Recommendations using Spark...
Retrieving Visually-Similar Products for Shopping Recommendations using Spark...Databricks
 

What's hot (20)

H2O World - Machine Learning at Comcast - Andrew Leamon & Chushi Ren
H2O World - Machine Learning at Comcast - Andrew Leamon & Chushi RenH2O World - Machine Learning at Comcast - Andrew Leamon & Chushi Ren
H2O World - Machine Learning at Comcast - Andrew Leamon & Chushi Ren
 
Dataiku hadoop summit - semi-supervised learning with hadoop for understand...
Dataiku   hadoop summit - semi-supervised learning with hadoop for understand...Dataiku   hadoop summit - semi-supervised learning with hadoop for understand...
Dataiku hadoop summit - semi-supervised learning with hadoop for understand...
 
Real-time Big Data Analytics: From Deployment to Production
Real-time Big Data Analytics: From Deployment to ProductionReal-time Big Data Analytics: From Deployment to Production
Real-time Big Data Analytics: From Deployment to Production
 
Leveraging Open Source Automated Data Science Tools
Leveraging Open Source Automated Data Science ToolsLeveraging Open Source Automated Data Science Tools
Leveraging Open Source Automated Data Science Tools
 
Danny Bickson - Python based predictive analytics with GraphLab Create
Danny Bickson - Python based predictive analytics with GraphLab Create Danny Bickson - Python based predictive analytics with GraphLab Create
Danny Bickson - Python based predictive analytics with GraphLab Create
 
Data Scientist Enablement roadmap 1.0
Data Scientist Enablement roadmap 1.0Data Scientist Enablement roadmap 1.0
Data Scientist Enablement roadmap 1.0
 
BigData
BigDataBigData
BigData
 
Using H2O for Mobile Transaction Forecasting & Anomaly Detection - Capital One
Using H2O for Mobile Transaction Forecasting & Anomaly Detection - Capital OneUsing H2O for Mobile Transaction Forecasting & Anomaly Detection - Capital One
Using H2O for Mobile Transaction Forecasting & Anomaly Detection - Capital One
 
Big Data, Baby Steps
Big Data, Baby StepsBig Data, Baby Steps
Big Data, Baby Steps
 
H2O for Medicine and Intro to H2O in Python
H2O for Medicine and Intro to H2O in PythonH2O for Medicine and Intro to H2O in Python
H2O for Medicine and Intro to H2O in Python
 
Introduction to Big Data: Smart Factory
Introduction to Big Data: Smart FactoryIntroduction to Big Data: Smart Factory
Introduction to Big Data: Smart Factory
 
From Machine Learning to Learning Machines: Creating an End-to-End Cognitive ...
From Machine Learning to Learning Machines: Creating an End-to-End Cognitive ...From Machine Learning to Learning Machines: Creating an End-to-End Cognitive ...
From Machine Learning to Learning Machines: Creating an End-to-End Cognitive ...
 
Data scientist enablement dse 400 week 8 roadmap
Data scientist enablement   dse 400   week 8 roadmap Data scientist enablement   dse 400   week 8 roadmap
Data scientist enablement dse 400 week 8 roadmap
 
Big Data and Predictive Analysis
Big Data and Predictive AnalysisBig Data and Predictive Analysis
Big Data and Predictive Analysis
 
Cloudera Breakfast: Advanced Analytics Part II: Do More With Your Data
Cloudera Breakfast: Advanced Analytics Part II: Do More With Your DataCloudera Breakfast: Advanced Analytics Part II: Do More With Your Data
Cloudera Breakfast: Advanced Analytics Part II: Do More With Your Data
 
Knowledge Discovery in Production
Knowledge Discovery in ProductionKnowledge Discovery in Production
Knowledge Discovery in Production
 
Agile Data Science by Russell Jurney_ The Hive_Janruary 29 2014
Agile Data Science by Russell Jurney_ The Hive_Janruary 29 2014Agile Data Science by Russell Jurney_ The Hive_Janruary 29 2014
Agile Data Science by Russell Jurney_ The Hive_Janruary 29 2014
 
AI on Big Data
AI on Big DataAI on Big Data
AI on Big Data
 
Data Science in the Real World: Making a Difference
Data Science in the Real World: Making a Difference Data Science in the Real World: Making a Difference
Data Science in the Real World: Making a Difference
 
Retrieving Visually-Similar Products for Shopping Recommendations using Spark...
Retrieving Visually-Similar Products for Shopping Recommendations using Spark...Retrieving Visually-Similar Products for Shopping Recommendations using Spark...
Retrieving Visually-Similar Products for Shopping Recommendations using Spark...
 

Similar to Dealing with uncertainty in fintech using AI

Flink Forward Berlin 2017: Bas Geerdink, Martijn Visser - Fast Data at ING - ...
Flink Forward Berlin 2017: Bas Geerdink, Martijn Visser - Fast Data at ING - ...Flink Forward Berlin 2017: Bas Geerdink, Martijn Visser - Fast Data at ING - ...
Flink Forward Berlin 2017: Bas Geerdink, Martijn Visser - Fast Data at ING - ...Flink Forward
 
Understanding Business APIs through statistics
Understanding Business APIs through statisticsUnderstanding Business APIs through statistics
Understanding Business APIs through statisticsWSO2
 
Accelerating ML using Production Feature Engineering
Accelerating ML using Production Feature EngineeringAccelerating ML using Production Feature Engineering
Accelerating ML using Production Feature EngineeringVenkata Pingali
 
Connecting Apache Kafka to Cash
Connecting Apache Kafka to CashConnecting Apache Kafka to Cash
Connecting Apache Kafka to Cashconfluent
 
Algolytics company Overview 2015
Algolytics company Overview 2015Algolytics company Overview 2015
Algolytics company Overview 2015Algolytics
 
Moving from BI to AI : For decision makers
Moving from BI to AI : For decision makersMoving from BI to AI : For decision makers
Moving from BI to AI : For decision makerszekeLabs Technologies
 
Data engineering in 10 years.pdf
Data engineering in 10 years.pdfData engineering in 10 years.pdf
Data engineering in 10 years.pdfLars Albertsson
 
Predicting Banking Customer Needs with an Agile Approach to Analytics in the ...
Predicting Banking Customer Needs with an Agile Approach to Analytics in the ...Predicting Banking Customer Needs with an Agile Approach to Analytics in the ...
Predicting Banking Customer Needs with an Agile Approach to Analytics in the ...Databricks
 
B2B Digital Transformation - Case Study
B2B Digital Transformation - Case StudyB2B Digital Transformation - Case Study
B2B Digital Transformation - Case StudyDivante
 
Ross Chayka. Business Models Generation
Ross Chayka. Business Models GenerationRoss Chayka. Business Models Generation
Ross Chayka. Business Models GenerationRostyslav Chayka
 
From Ideation to Production in 7 days: The Scoring Factory at Raiffeisen
From Ideation to Production in 7 days: The Scoring Factory at RaiffeisenFrom Ideation to Production in 7 days: The Scoring Factory at Raiffeisen
From Ideation to Production in 7 days: The Scoring Factory at RaiffeisenBATbern
 
EY + Neo4j: Why graph technology makes sense for fraud detection and customer...
EY + Neo4j: Why graph technology makes sense for fraud detection and customer...EY + Neo4j: Why graph technology makes sense for fraud detection and customer...
EY + Neo4j: Why graph technology makes sense for fraud detection and customer...Neo4j
 
Making machine learning model deployment boring - Big Data Expo 2019
Making machine learning model deployment boring - Big Data Expo 2019Making machine learning model deployment boring - Big Data Expo 2019
Making machine learning model deployment boring - Big Data Expo 2019webwinkelvakdag
 
Emvigo Data Visualization - E Commerce Deck
Emvigo Data Visualization - E Commerce DeckEmvigo Data Visualization - E Commerce Deck
Emvigo Data Visualization - E Commerce DeckEmvigo Technologies
 
Understanding Computers: Today and Tomorrow, 13th Edition Chapter 11 - E-Comm...
Understanding Computers: Today and Tomorrow, 13th Edition Chapter 11 - E-Comm...Understanding Computers: Today and Tomorrow, 13th Edition Chapter 11 - E-Comm...
Understanding Computers: Today and Tomorrow, 13th Edition Chapter 11 - E-Comm...yaminohime
 
SaaS startups - Software Engineering Challenges
SaaS startups - Software Engineering ChallengesSaaS startups - Software Engineering Challenges
SaaS startups - Software Engineering ChallengesMalinda Kapuruge
 
[APIdays Singapore 2019] Implementing a Successful Open Banking Architecture
[APIdays Singapore 2019] Implementing a Successful Open Banking Architecture[APIdays Singapore 2019] Implementing a Successful Open Banking Architecture
[APIdays Singapore 2019] Implementing a Successful Open Banking ArchitectureWSO2
 

Similar to Dealing with uncertainty in fintech using AI (20)

Real-Time Machine Learning at Industrial scale (University of Oxford, 9th Oct...
Real-Time Machine Learning at Industrial scale (University of Oxford, 9th Oct...Real-Time Machine Learning at Industrial scale (University of Oxford, 9th Oct...
Real-Time Machine Learning at Industrial scale (University of Oxford, 9th Oct...
 
Flink Forward Berlin 2017: Bas Geerdink, Martijn Visser - Fast Data at ING - ...
Flink Forward Berlin 2017: Bas Geerdink, Martijn Visser - Fast Data at ING - ...Flink Forward Berlin 2017: Bas Geerdink, Martijn Visser - Fast Data at ING - ...
Flink Forward Berlin 2017: Bas Geerdink, Martijn Visser - Fast Data at ING - ...
 
Understanding Business APIs through statistics
Understanding Business APIs through statisticsUnderstanding Business APIs through statistics
Understanding Business APIs through statistics
 
Accelerating ML using Production Feature Engineering
Accelerating ML using Production Feature EngineeringAccelerating ML using Production Feature Engineering
Accelerating ML using Production Feature Engineering
 
Connecting Apache Kafka to Cash
Connecting Apache Kafka to CashConnecting Apache Kafka to Cash
Connecting Apache Kafka to Cash
 
Algolytics company Overview 2015
Algolytics company Overview 2015Algolytics company Overview 2015
Algolytics company Overview 2015
 
Algolytics company Overview 2015
Algolytics company Overview 2015Algolytics company Overview 2015
Algolytics company Overview 2015
 
Moving from BI to AI : For decision makers
Moving from BI to AI : For decision makersMoving from BI to AI : For decision makers
Moving from BI to AI : For decision makers
 
Tandem Bank and AWS
Tandem Bank and AWSTandem Bank and AWS
Tandem Bank and AWS
 
Data engineering in 10 years.pdf
Data engineering in 10 years.pdfData engineering in 10 years.pdf
Data engineering in 10 years.pdf
 
Predicting Banking Customer Needs with an Agile Approach to Analytics in the ...
Predicting Banking Customer Needs with an Agile Approach to Analytics in the ...Predicting Banking Customer Needs with an Agile Approach to Analytics in the ...
Predicting Banking Customer Needs with an Agile Approach to Analytics in the ...
 
B2B Digital Transformation - Case Study
B2B Digital Transformation - Case StudyB2B Digital Transformation - Case Study
B2B Digital Transformation - Case Study
 
Ross Chayka. Business Models Generation
Ross Chayka. Business Models GenerationRoss Chayka. Business Models Generation
Ross Chayka. Business Models Generation
 
From Ideation to Production in 7 days: The Scoring Factory at Raiffeisen
From Ideation to Production in 7 days: The Scoring Factory at RaiffeisenFrom Ideation to Production in 7 days: The Scoring Factory at Raiffeisen
From Ideation to Production in 7 days: The Scoring Factory at Raiffeisen
 
EY + Neo4j: Why graph technology makes sense for fraud detection and customer...
EY + Neo4j: Why graph technology makes sense for fraud detection and customer...EY + Neo4j: Why graph technology makes sense for fraud detection and customer...
EY + Neo4j: Why graph technology makes sense for fraud detection and customer...
 
Making machine learning model deployment boring - Big Data Expo 2019
Making machine learning model deployment boring - Big Data Expo 2019Making machine learning model deployment boring - Big Data Expo 2019
Making machine learning model deployment boring - Big Data Expo 2019
 
Emvigo Data Visualization - E Commerce Deck
Emvigo Data Visualization - E Commerce DeckEmvigo Data Visualization - E Commerce Deck
Emvigo Data Visualization - E Commerce Deck
 
Understanding Computers: Today and Tomorrow, 13th Edition Chapter 11 - E-Comm...
Understanding Computers: Today and Tomorrow, 13th Edition Chapter 11 - E-Comm...Understanding Computers: Today and Tomorrow, 13th Edition Chapter 11 - E-Comm...
Understanding Computers: Today and Tomorrow, 13th Edition Chapter 11 - E-Comm...
 
SaaS startups - Software Engineering Challenges
SaaS startups - Software Engineering ChallengesSaaS startups - Software Engineering Challenges
SaaS startups - Software Engineering Challenges
 
[APIdays Singapore 2019] Implementing a Successful Open Banking Architecture
[APIdays Singapore 2019] Implementing a Successful Open Banking Architecture[APIdays Singapore 2019] Implementing a Successful Open Banking Architecture
[APIdays Singapore 2019] Implementing a Successful Open Banking Architecture
 

More from Data Products Meetup

"Chat-bots and AI: the reality and misconceptions" by Vyugar Nadzhafov
"Chat-bots and AI: the reality and misconceptions" by Vyugar Nadzhafov"Chat-bots and AI: the reality and misconceptions" by Vyugar Nadzhafov
"Chat-bots and AI: the reality and misconceptions" by Vyugar NadzhafovData Products Meetup
 
"How did I learn how not to run a Data Science project" by Florian Bonnet
"How did I learn how not to run a Data Science project" by Florian Bonnet"How did I learn how not to run a Data Science project" by Florian Bonnet
"How did I learn how not to run a Data Science project" by Florian BonnetData Products Meetup
 
Classifications as Power Technologies by Milagros Miceli
Classifications as Power Technologies by Milagros MiceliClassifications as Power Technologies by Milagros Miceli
Classifications as Power Technologies by Milagros MiceliData Products Meetup
 
"8 donts and dos when building a chatbot" by Yura Okulovsky
"8 donts and dos when building a chatbot" by Yura Okulovsky"8 donts and dos when building a chatbot" by Yura Okulovsky
"8 donts and dos when building a chatbot" by Yura OkulovskyData Products Meetup
 
The ways to fuck up ab testing (from data products meetup)
The ways to fuck up ab testing (from data products meetup)The ways to fuck up ab testing (from data products meetup)
The ways to fuck up ab testing (from data products meetup)Data Products Meetup
 

More from Data Products Meetup (7)

"Chat-bots and AI: the reality and misconceptions" by Vyugar Nadzhafov
"Chat-bots and AI: the reality and misconceptions" by Vyugar Nadzhafov"Chat-bots and AI: the reality and misconceptions" by Vyugar Nadzhafov
"Chat-bots and AI: the reality and misconceptions" by Vyugar Nadzhafov
 
"How did I learn how not to run a Data Science project" by Florian Bonnet
"How did I learn how not to run a Data Science project" by Florian Bonnet"How did I learn how not to run a Data Science project" by Florian Bonnet
"How did I learn how not to run a Data Science project" by Florian Bonnet
 
Classifications as Power Technologies by Milagros Miceli
Classifications as Power Technologies by Milagros MiceliClassifications as Power Technologies by Milagros Miceli
Classifications as Power Technologies by Milagros Miceli
 
"8 donts and dos when building a chatbot" by Yura Okulovsky
"8 donts and dos when building a chatbot" by Yura Okulovsky"8 donts and dos when building a chatbot" by Yura Okulovsky
"8 donts and dos when building a chatbot" by Yura Okulovsky
 
Racial and gender bias in AI
Racial and gender bias in AI Racial and gender bias in AI
Racial and gender bias in AI
 
Quantifying fashion
Quantifying fashionQuantifying fashion
Quantifying fashion
 
The ways to fuck up ab testing (from data products meetup)
The ways to fuck up ab testing (from data products meetup)The ways to fuck up ab testing (from data products meetup)
The ways to fuck up ab testing (from data products meetup)
 

Recently uploaded

Welcome to GFDL for Take Your Child To Work Day
Welcome to GFDL for Take Your Child To Work DayWelcome to GFDL for Take Your Child To Work Day
Welcome to GFDL for Take Your Child To Work DayZachary Labe
 
Temporomandibular joint Muscles of Mastication
Temporomandibular joint Muscles of MasticationTemporomandibular joint Muscles of Mastication
Temporomandibular joint Muscles of Masticationvidulajaib
 
Behavioral Disorder: Schizophrenia & it's Case Study.pdf
Behavioral Disorder: Schizophrenia & it's Case Study.pdfBehavioral Disorder: Schizophrenia & it's Case Study.pdf
Behavioral Disorder: Schizophrenia & it's Case Study.pdfSELF-EXPLANATORY
 
Recombinant DNA technology( Transgenic plant and animal)
Recombinant DNA technology( Transgenic plant and animal)Recombinant DNA technology( Transgenic plant and animal)
Recombinant DNA technology( Transgenic plant and animal)DHURKADEVIBASKAR
 
Call Girls In Nihal Vihar Delhi ❤️8860477959 Looking Escorts In 24/7 Delhi NCR
Call Girls In Nihal Vihar Delhi ❤️8860477959 Looking Escorts In 24/7 Delhi NCRCall Girls In Nihal Vihar Delhi ❤️8860477959 Looking Escorts In 24/7 Delhi NCR
Call Girls In Nihal Vihar Delhi ❤️8860477959 Looking Escorts In 24/7 Delhi NCRlizamodels9
 
TOPIC 8 Temperature and Heat.pdf physics
TOPIC 8 Temperature and Heat.pdf physicsTOPIC 8 Temperature and Heat.pdf physics
TOPIC 8 Temperature and Heat.pdf physicsssuserddc89b
 
Analytical Profile of Coleus Forskohlii | Forskolin .pdf
Analytical Profile of Coleus Forskohlii | Forskolin .pdfAnalytical Profile of Coleus Forskohlii | Forskolin .pdf
Analytical Profile of Coleus Forskohlii | Forskolin .pdfSwapnil Therkar
 
Transposable elements in prokaryotes.ppt
Transposable elements in prokaryotes.pptTransposable elements in prokaryotes.ppt
Transposable elements in prokaryotes.pptArshadWarsi13
 
Grafana in space: Monitoring Japan's SLIM moon lander in real time
Grafana in space: Monitoring Japan's SLIM moon lander  in real timeGrafana in space: Monitoring Japan's SLIM moon lander  in real time
Grafana in space: Monitoring Japan's SLIM moon lander in real timeSatoshi NAKAHIRA
 
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝soniya singh
 
THE ROLE OF PHARMACOGNOSY IN TRADITIONAL AND MODERN SYSTEM OF MEDICINE.pptx
THE ROLE OF PHARMACOGNOSY IN TRADITIONAL AND MODERN SYSTEM OF MEDICINE.pptxTHE ROLE OF PHARMACOGNOSY IN TRADITIONAL AND MODERN SYSTEM OF MEDICINE.pptx
THE ROLE OF PHARMACOGNOSY IN TRADITIONAL AND MODERN SYSTEM OF MEDICINE.pptxNandakishor Bhaurao Deshmukh
 
TOTAL CHOLESTEROL (lipid profile test).pptx
TOTAL CHOLESTEROL (lipid profile test).pptxTOTAL CHOLESTEROL (lipid profile test).pptx
TOTAL CHOLESTEROL (lipid profile test).pptxdharshini369nike
 
LIGHT-PHENOMENA-BY-CABUALDIONALDOPANOGANCADIENTE-CONDEZA (1).pptx
LIGHT-PHENOMENA-BY-CABUALDIONALDOPANOGANCADIENTE-CONDEZA (1).pptxLIGHT-PHENOMENA-BY-CABUALDIONALDOPANOGANCADIENTE-CONDEZA (1).pptx
LIGHT-PHENOMENA-BY-CABUALDIONALDOPANOGANCADIENTE-CONDEZA (1).pptxmalonesandreagweneth
 
Bentham & Hooker's Classification. along with the merits and demerits of the ...
Bentham & Hooker's Classification. along with the merits and demerits of the ...Bentham & Hooker's Classification. along with the merits and demerits of the ...
Bentham & Hooker's Classification. along with the merits and demerits of the ...Nistarini College, Purulia (W.B) India
 
insect anatomy and insect body wall and their physiology
insect anatomy and insect body wall and their  physiologyinsect anatomy and insect body wall and their  physiology
insect anatomy and insect body wall and their physiologyDrAnita Sharma
 
Harmful and Useful Microorganisms Presentation
Harmful and Useful Microorganisms PresentationHarmful and Useful Microorganisms Presentation
Harmful and Useful Microorganisms Presentationtahreemzahra82
 
Vision and reflection on Mining Software Repositories research in 2024
Vision and reflection on Mining Software Repositories research in 2024Vision and reflection on Mining Software Repositories research in 2024
Vision and reflection on Mining Software Repositories research in 2024AyushiRastogi48
 
Spermiogenesis or Spermateleosis or metamorphosis of spermatid
Spermiogenesis or Spermateleosis or metamorphosis of spermatidSpermiogenesis or Spermateleosis or metamorphosis of spermatid
Spermiogenesis or Spermateleosis or metamorphosis of spermatidSarthak Sekhar Mondal
 
Speech, hearing, noise, intelligibility.pptx
Speech, hearing, noise, intelligibility.pptxSpeech, hearing, noise, intelligibility.pptx
Speech, hearing, noise, intelligibility.pptxpriyankatabhane
 

Recently uploaded (20)

Welcome to GFDL for Take Your Child To Work Day
Welcome to GFDL for Take Your Child To Work DayWelcome to GFDL for Take Your Child To Work Day
Welcome to GFDL for Take Your Child To Work Day
 
Temporomandibular joint Muscles of Mastication
Temporomandibular joint Muscles of MasticationTemporomandibular joint Muscles of Mastication
Temporomandibular joint Muscles of Mastication
 
Behavioral Disorder: Schizophrenia & it's Case Study.pdf
Behavioral Disorder: Schizophrenia & it's Case Study.pdfBehavioral Disorder: Schizophrenia & it's Case Study.pdf
Behavioral Disorder: Schizophrenia & it's Case Study.pdf
 
Recombinant DNA technology( Transgenic plant and animal)
Recombinant DNA technology( Transgenic plant and animal)Recombinant DNA technology( Transgenic plant and animal)
Recombinant DNA technology( Transgenic plant and animal)
 
Volatile Oils Pharmacognosy And Phytochemistry -I
Volatile Oils Pharmacognosy And Phytochemistry -IVolatile Oils Pharmacognosy And Phytochemistry -I
Volatile Oils Pharmacognosy And Phytochemistry -I
 
Call Girls In Nihal Vihar Delhi ❤️8860477959 Looking Escorts In 24/7 Delhi NCR
Call Girls In Nihal Vihar Delhi ❤️8860477959 Looking Escorts In 24/7 Delhi NCRCall Girls In Nihal Vihar Delhi ❤️8860477959 Looking Escorts In 24/7 Delhi NCR
Call Girls In Nihal Vihar Delhi ❤️8860477959 Looking Escorts In 24/7 Delhi NCR
 
TOPIC 8 Temperature and Heat.pdf physics
TOPIC 8 Temperature and Heat.pdf physicsTOPIC 8 Temperature and Heat.pdf physics
TOPIC 8 Temperature and Heat.pdf physics
 
Analytical Profile of Coleus Forskohlii | Forskolin .pdf
Analytical Profile of Coleus Forskohlii | Forskolin .pdfAnalytical Profile of Coleus Forskohlii | Forskolin .pdf
Analytical Profile of Coleus Forskohlii | Forskolin .pdf
 
Transposable elements in prokaryotes.ppt
Transposable elements in prokaryotes.pptTransposable elements in prokaryotes.ppt
Transposable elements in prokaryotes.ppt
 
Grafana in space: Monitoring Japan's SLIM moon lander in real time
Grafana in space: Monitoring Japan's SLIM moon lander  in real timeGrafana in space: Monitoring Japan's SLIM moon lander  in real time
Grafana in space: Monitoring Japan's SLIM moon lander in real time
 
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝
 
THE ROLE OF PHARMACOGNOSY IN TRADITIONAL AND MODERN SYSTEM OF MEDICINE.pptx
THE ROLE OF PHARMACOGNOSY IN TRADITIONAL AND MODERN SYSTEM OF MEDICINE.pptxTHE ROLE OF PHARMACOGNOSY IN TRADITIONAL AND MODERN SYSTEM OF MEDICINE.pptx
THE ROLE OF PHARMACOGNOSY IN TRADITIONAL AND MODERN SYSTEM OF MEDICINE.pptx
 
TOTAL CHOLESTEROL (lipid profile test).pptx
TOTAL CHOLESTEROL (lipid profile test).pptxTOTAL CHOLESTEROL (lipid profile test).pptx
TOTAL CHOLESTEROL (lipid profile test).pptx
 
LIGHT-PHENOMENA-BY-CABUALDIONALDOPANOGANCADIENTE-CONDEZA (1).pptx
LIGHT-PHENOMENA-BY-CABUALDIONALDOPANOGANCADIENTE-CONDEZA (1).pptxLIGHT-PHENOMENA-BY-CABUALDIONALDOPANOGANCADIENTE-CONDEZA (1).pptx
LIGHT-PHENOMENA-BY-CABUALDIONALDOPANOGANCADIENTE-CONDEZA (1).pptx
 
Bentham & Hooker's Classification. along with the merits and demerits of the ...
Bentham & Hooker's Classification. along with the merits and demerits of the ...Bentham & Hooker's Classification. along with the merits and demerits of the ...
Bentham & Hooker's Classification. along with the merits and demerits of the ...
 
insect anatomy and insect body wall and their physiology
insect anatomy and insect body wall and their  physiologyinsect anatomy and insect body wall and their  physiology
insect anatomy and insect body wall and their physiology
 
Harmful and Useful Microorganisms Presentation
Harmful and Useful Microorganisms PresentationHarmful and Useful Microorganisms Presentation
Harmful and Useful Microorganisms Presentation
 
Vision and reflection on Mining Software Repositories research in 2024
Vision and reflection on Mining Software Repositories research in 2024Vision and reflection on Mining Software Repositories research in 2024
Vision and reflection on Mining Software Repositories research in 2024
 
Spermiogenesis or Spermateleosis or metamorphosis of spermatid
Spermiogenesis or Spermateleosis or metamorphosis of spermatidSpermiogenesis or Spermateleosis or metamorphosis of spermatid
Spermiogenesis or Spermateleosis or metamorphosis of spermatid
 
Speech, hearing, noise, intelligibility.pptx
Speech, hearing, noise, intelligibility.pptxSpeech, hearing, noise, intelligibility.pptx
Speech, hearing, noise, intelligibility.pptx
 

Dealing with uncertainty in fintech using AI

  • 1. 1
  • 2. 01 Table of contents Smava business in a nutshell 02 Why ML and AI? 03 Production environment 04 Deploying to AWS 05 Summary 2
  • 4. 4 A comparison platform for personal loans 1. 1000-50000 Euros,expanding into larger size loans (i.e. mortgage). 2. Financing for purchasing a car, repairing your home, travelling etc. 3. Provided by one of the partner banks 4. Needs to be paid back ○ Customer’s creditworthiness ○ Fundamentally different to a typical ‘buy and forget’ comparison platform
  • 5. 5 Business Model smava enables borrowers to get the best deal for consumer loans, by pooling offers from a few dozen banks in a single website
  • 6. 6 Customer journey 1. www.smava.de - fill in the details online ○ An average customer has to answer 50 questions ~ 15 mins 2. Smava sends an API request to partner banks and gets a response ~2 mins ○ As soon as the registration route is finished partner banks are requested and offers are displayed 3. Provide proof of income and other details ~ 7 days ○ Usually scans of income slips, bank statements, etc. 4. Get the money into your bank account or get a rejection.
  • 7. 7
  • 8. 8
  • 9. Why ML and AI? 02 9
  • 10. 10 Problem statement ● Banks have varying requirements and preferences reflected by their risk management policy ○ Minimise defaults, maximise ROI ● Banks incur costs when assessing leads from Smava ● Customer needs to qualify for the loan ○ How to efficiently match between customer and a bank? ● Customer only cares about the cost of the loan and the ease of getting it
  • 11. Why machine learning? ● Uncertainty - what is the probability that customer x will get accepted by bank y? ● Ignore uncertainty ○ request all banks for each individual customer ○ Banks don’t like waste (money spent on leads that did not convert) ● Address uncertainty with heuristics ○ High income customers - Bank A, low income customers - Bank B ○ Not efficient, needs continuous updating and manual tweaking ○ Could potentially be more harmful to the business than ignoring uncertainty ● Address uncertainty with Machine Learning ○ Use the full set of data features ■ To learn and predict bank answers ○ Retrain models frequently ○ Use predictions for efficient matching (how?!) ● Blend of heuristics and Machine Learning ○ Heuristics were there first and are believed to be useful ○ Some very simple rules can be useful 11
  • 12. smava Master Presentation Template - Introduction 12 Smava marketplace ● Efficient matching ○ Customer mostly care about the interest rate ○ Banks care about number of requests they receive vs the number of payouts they provide. Banks do not like waste ○ smava’s mission is to assist the customer find the best deal on the marketplace ■ smava is concerned both with the interest rate of the offer AND the realisability of the offer (measured by probability to get the offer)
  • 14. Smava engineering infrastructure 14 Data collection, processing saving (Smava website, databases etc.) Predictions, portfolio creation, execution schema (Machine Learning microservice) Bank API connectivity Bank_1 Bank_2 Bank_N customer data request execution schema offer request - response offer request - response ~15 mins ~1 sec ~60 sec t
  • 15. Design principles ● Close collaboration with engineering ○ Impact on all major KPIs ● Extensive tests ○ Unit tests ○ Statistical tests ● Automatic retraining - weekly/daily ● Live monitoring ○ input data ○ output data 15
  • 16. Automatic retraining ● Get data from all sources ● Clean data and prepare it for modelling ● Train models ● Run tests ● Compute predictive accuracy on unseen data ● Return a retraining report 16
  • 17. Monitoring ● Real time model predictive performance monitoring - are we predicting well? ● Anomaly detection - has anything in the outside world changed? 17
  • 18. 18 Tech stack ● Python for data pipelines ● R for modelling and production code ● Shiny for monitoring ● C for some internal packages
  • 19. 19 Predicting and using the predictions ● XGboost for predicting probabilities ● Using probabilities to guide decision making ● Intellectual property
  • 20. Model hosting on AWS 04 20
  • 21. 21 ● AWS Sagemker: platform to deploy ML model in cloud ○ Competitors: Azure AI, Google Cloud AI ● Fully managed ○ autoscaling, health checks, load balancing ○ allows to change types and number of instances; no downtime ● Can deploy custom Docker image ● Next iteration : AWS Lambda Model deployment
  • 22. 22 ● Dockerized REST API (using Nginx, Gunicorn and Flask) ● SM expects server to respond to /invocations and /ping (port 8080) ○ /invocations: accept POST requests and returns prediction ○ /ping: accept GET and returns 200 (for health checks) Model deployment
  • 23. 23 23 23 ● Amazon cloudwatch: ○ latency, invocations, cpu-utilization,.. Monitoring
  • 24. 24 24 ● Create a new endpoint and put in shadow for a day ● Live monitoring of shadow Blue-Green deployment Prod Shadow S3 Further Processing Input output output
  • 25. 25 ● DS team fully owns the Application Engine microservice ○ Complete ownership of infrastructure ○ DS outputs are provided via a REST api Application Engine microservice
  • 27. 27 ● ML component within smava is the very core of the production system ● Opportunities to significantly impact the business in real time ○ Big potential for both improvement, but also for causing damage ● Hand in hand work with engineering, product owners and wider stakeholders in the company ○ Advocacy ○ Explanations ○ Conflicts of interest Summary
  • 28. Thank you for your attention Evgeny Savin Senior data scientists 28 evgeny.savin@smava.de