SlideShare a Scribd company logo
1 of 28
Project Management
in AI
Nathanael Weill, PhD
About me
2
Master in Bioinformatics Strasbourg University (France)
Ph.D. In Pharmaceutical Science. Strasbourg University (France)
Post-Doc at McGill (Computational chemistry)
Post-Doc at UdeM (Computational Biology)
Senior Data Scientist at Mnubo (IoT company)
Nathanael Weill
What is AI?
Why AI?
AI project phases
Warnings
Optimize the process
Outlines
3
The theory and development of computer systems able to perform
tasks that normally require human intelligence, such as visual perception,
speech recognition, decision-making, and translation between languages. (google
dictionary)
What is AI?
4
Prediction: The process of filling in
missing information. Prediction takes data
you have to generate data you don’t have.
How does it work?
5
computer
Input data
Output
Function
computer
Input data
Function
Output
computer
New Input data
Prediction
Function
Why AI?
6
Prediction became cheaper
Data
AccuracyClient
The AI race
7
Big Data & Data Science Projects
Failure Rate
9
GARTNER
ESTIMATED
85%
of big data projects
fail (2017). The
initial estimation
was 60% (GARTNER
2016)
THROUGH 2020
80%
of AI projects will
remain alchemy,
run by wizards
whose talents will
not scale in the
organization.
(GARTNER 2018)
THROUGH 2022
20%
of analytic insights
will deliver
business
outcomes. (GARTNER
2018)
EXECUTIVE
SURVEY
77%
respondents say
that “business
adoption” of big
data and AI
initiatives
continues to
represent a
challenge for their
organizations
(NEWVANTAGE
PARTNERS 2019)
A recipe for failure
We must define the solution as an entire process.
If prediction is the end of the solution, the entire solution might fail because:
• The output does not correspond to the operational needs.
• The operator will not use it due to complexification of the process.
• No one is capable of managing the algorithms if something goes wrong.
• …
Data Algorithm Prediction
Data Algorithm Prediction
Judgment
Action
Feedback
Critical! We have to make sure we produce the right
information and in the right format to help the person in charge
to take action
Manager: Person in charge to take action. We need to
make sure this person is involved early in the process
Design of the solution
Identification of
the problem to
solve
Design the
appropriate
solution
Proof of concept
Productization
Scale the process
Reorganize the
company
6 Phases
12
Use Case
13
At Mnubo we designed a 3-5 days workshop
with clients to go from the problem identification
to the mock up of the solution
Performance problem? Scalability issue?
How to Consume the predictions? Maintain the solution?
What action(s) will be taken?
…
Ex:
1 prediction per machine? Every hour? 12 hours?
Solving the right problem
14
A journey as a Data Scientist 1/2
15
Data Scientist:
Define the valuable business
problem
Translate the business problem
into a KPI
A Key Performance Indicator (KPI) is a
measurable value that demonstrates how
effectively a company is achieving key
business objectives. Organizations use key
performance indicators at multiple levels to
evaluate their success at reaching targets.
Client:
« I loose a lot of money when the
assembly lines stops ».
« I would like to reduce the number of
machine failures ».
https://www.klipfolio.com/resources/kpi-examples
A journey as a Data Scientist 2/2
16
Data Scientist:
Define the metric and the
definition of success.
Next phase: Proof of concept.
• explore
• Establish a baseline
• Iterate!!!
Client:
A success would be to predict
failure 12 hours in advance
with an accuracy of 80%
According to the final report, I
get an answer to:
• Is the objectives reasonable?
• How should I productize the
solution?
POC: Critical choice
17
Time
Resources Accuracy
• Explore
• Create a baseline
• Iterate
Agile
Productization phase
18
2 productization models:
• Data scientist write specifications and engineers take over and
rewrite the code in an other language (java, scala…)
• Data scientist with a team of data engineer, dev ops etc… takes
the code written and deploy it in the infrastructure
Pros and cons…
Data Algorithm Prediction
Judgment
Action
Feedback
Data Algorithm Prediction
Judgment
Action
Feedback
Data Algorithm Prediction
Judgment
Action
Feedback
Data Algorithm Prediction
Judgment
Action
Feedback
Data Algorithm Prediction
Judgment
Action
Feedback
Data Algorithm Prediction
Judgment
Action
Feedback
Data Algorithm Prediction
Judgment
Action
Feedback
Full solution management:
• Configuration
• Monitoring
• ROI evaluation
Scaling of the Solution
Avoid silos
labyrinthine system
Data Algorithm
Judgment
Action
Feedback
Data Algorithm Prediction
Judgment
Action
Feedback
(Automating)
Dev ops: In charge of deploying and maintaining the
infrastructure to support the solution
Data engineer: in charge of setting the appropriate
resource to access the data.
Data scientist: in charge of creating the machine
learning model (pipeline data to prediction)
Roles: development phases
21
Operator: In charge of activating/deactivating the
algorithms designed for specific predictions/actions
=> Provide feedback to data scientists
Data scientist: Integrate the feedback and update the
algorithm (if needed)
Dev ops: Maintain the infrastructure
Roles: long term
22
Company perturbation
23
IT Team
Operation
Team
Executives
Data Science
Team
24
The Proof of Concept Curse in AI and IoT
80% of companies stop at
the POC stage.
Laggards & Winners
I recommend:
To use Agile methodology in all
phases of the project
Have a clear understanding of
the final aim in term of:
• The process of development
• The perturbation of the company
organization
Critical role of the project manager
25
Phases:
Identification of the problem to
solve
Design the appropriate solution
Proof of concept
Productization
Scale the process
Reorganize the company
There is multiple tracks that can be done in parallel:
Data acquisition
To make sure the data are available in (near-) real time.
Creation of the machine learning algorithm
Create the appropriate pipeline to train, test and deploy the model(s).
Creation of the end point to expose the predictions
A dashboard, an app, an alerting system, a reporting system.
Monitoring of the pipeline
monitor the data acquisition, the performance of the model, the use of the
end point…
Process to capture the action taken and consolidate a feedback
loop
Optimize the process
26
Hofstadter's law: It always takes longer than you expect, even when
you take into account Hofstadter's Law.
First AI project is hard, you should start with an easy project
• Is there already a system in place to monitor the KPI?
• Is the data pipeline already in place?
• Is AI a replacement for an existing system?
Assess the client maturity is hard especially regarding the company
perturbation
A good PM is the key to success!
Wrap up
27
Nathanael Weill
nweill@mnubo.com
28
Thank you!!

More Related Content

What's hot

Principles of Artificial Intelligence & Machine Learning
Principles of Artificial Intelligence & Machine LearningPrinciples of Artificial Intelligence & Machine Learning
Principles of Artificial Intelligence & Machine Learning
Jerry Lu
 
Artificial intelligence-full -report.doc
Artificial intelligence-full -report.docArtificial intelligence-full -report.doc
Artificial intelligence-full -report.doc
daksh Talsaniya
 

What's hot (20)

From Narrow AI to Artificial General Intelligence (AGI)
From Narrow AI to Artificial General Intelligence (AGI)From Narrow AI to Artificial General Intelligence (AGI)
From Narrow AI to Artificial General Intelligence (AGI)
 
L’optimisation par essaims de particules
L’optimisation par essaims de particulesL’optimisation par essaims de particules
L’optimisation par essaims de particules
 
Artificial intelligence
Artificial intelligenceArtificial intelligence
Artificial intelligence
 
Artificial Intelligence (A.I.) || Introduction of A.I. || HELPFUL FOR STUDENT...
Artificial Intelligence (A.I.) || Introduction of A.I. || HELPFUL FOR STUDENT...Artificial Intelligence (A.I.) || Introduction of A.I. || HELPFUL FOR STUDENT...
Artificial Intelligence (A.I.) || Introduction of A.I. || HELPFUL FOR STUDENT...
 
Principles of Artificial Intelligence & Machine Learning
Principles of Artificial Intelligence & Machine LearningPrinciples of Artificial Intelligence & Machine Learning
Principles of Artificial Intelligence & Machine Learning
 
Artificial intelligence-full -report.doc
Artificial intelligence-full -report.docArtificial intelligence-full -report.doc
Artificial intelligence-full -report.doc
 
Presentation PFE
Presentation PFEPresentation PFE
Presentation PFE
 
Artificial Intelligence
Artificial IntelligenceArtificial Intelligence
Artificial Intelligence
 
BLUE BRAIN
BLUE BRAINBLUE BRAIN
BLUE BRAIN
 
la 2éme édition de notre journal scolaire. ' collège d'akka'. d'après: Zahour...
la 2éme édition de notre journal scolaire. ' collège d'akka'. d'après: Zahour...la 2éme édition de notre journal scolaire. ' collège d'akka'. d'après: Zahour...
la 2éme édition de notre journal scolaire. ' collège d'akka'. d'après: Zahour...
 
Artificial intelligence
Artificial intelligenceArtificial intelligence
Artificial intelligence
 
blue brain
blue brainblue brain
blue brain
 
Ai for everyone
Ai for everyoneAi for everyone
Ai for everyone
 
Artificial Intellegence
Artificial IntellegenceArtificial Intellegence
Artificial Intellegence
 
AI Math Agents
AI Math AgentsAI Math Agents
AI Math Agents
 
5 PEN PC TECHNOLOGY with GEN OF Computers
5 PEN PC TECHNOLOGY with GEN OF Computers5 PEN PC TECHNOLOGY with GEN OF Computers
5 PEN PC TECHNOLOGY with GEN OF Computers
 
Blue brain project
Blue brain projectBlue brain project
Blue brain project
 
5 pen-pc-Technology
5 pen-pc-Technology5 pen-pc-Technology
5 pen-pc-Technology
 
Introduction to soft computing
Introduction to soft computingIntroduction to soft computing
Introduction to soft computing
 
AI_1 Introduction of AI
AI_1 Introduction of AIAI_1 Introduction of AI
AI_1 Introduction of AI
 

Similar to Symposium 2019 : Gestion de projet en Intelligence Artificielle

Agile Mumbai 2023 | AI-Powered Agility: A New Era of Sustainable Business Inn...
Agile Mumbai 2023 | AI-Powered Agility: A New Era of Sustainable Business Inn...Agile Mumbai 2023 | AI-Powered Agility: A New Era of Sustainable Business Inn...
Agile Mumbai 2023 | AI-Powered Agility: A New Era of Sustainable Business Inn...
AgileNetwork
 
Challenges of Executing AI
Challenges of Executing AIChallenges of Executing AI
Challenges of Executing AI
Dr. Umesh Rao.Hodeghatta
 
integrating-cognitive-services-into-your-devops-strategy
integrating-cognitive-services-into-your-devops-strategyintegrating-cognitive-services-into-your-devops-strategy
integrating-cognitive-services-into-your-devops-strategy
Karthik Jaganathan
 

Similar to Symposium 2019 : Gestion de projet en Intelligence Artificielle (20)

[DSC Europe 22] The Making of a Data Organization - Denys Holovatyi
[DSC Europe 22] The Making of a Data Organization - Denys Holovatyi[DSC Europe 22] The Making of a Data Organization - Denys Holovatyi
[DSC Europe 22] The Making of a Data Organization - Denys Holovatyi
 
How to make your data scientists happy
How to make your data scientists happy How to make your data scientists happy
How to make your data scientists happy
 
Data science is not Software Development and how Experiment Management can ma...
Data science is not Software Development and how Experiment Management can ma...Data science is not Software Development and how Experiment Management can ma...
Data science is not Software Development and how Experiment Management can ma...
 
Agile Mumbai 2023 | AI-Powered Agility: A New Era of Sustainable Business Inn...
Agile Mumbai 2023 | AI-Powered Agility: A New Era of Sustainable Business Inn...Agile Mumbai 2023 | AI-Powered Agility: A New Era of Sustainable Business Inn...
Agile Mumbai 2023 | AI-Powered Agility: A New Era of Sustainable Business Inn...
 
Intelligently Automating Machine Learning, Artificial Intelligence, and Data ...
Intelligently Automating Machine Learning, Artificial Intelligence, and Data ...Intelligently Automating Machine Learning, Artificial Intelligence, and Data ...
Intelligently Automating Machine Learning, Artificial Intelligence, and Data ...
 
Challenges of Executing AI
Challenges of Executing AIChallenges of Executing AI
Challenges of Executing AI
 
Big Data LDN 2018: THE PATH TO ENTERPRISE AI: TALES FROM THE FIELD
Big Data LDN 2018: THE PATH TO ENTERPRISE AI: TALES FROM THE FIELDBig Data LDN 2018: THE PATH TO ENTERPRISE AI: TALES FROM THE FIELD
Big Data LDN 2018: THE PATH TO ENTERPRISE AI: TALES FROM THE FIELD
 
Metrics for Mofel-Based Systems Development
Metrics for Mofel-Based Systems DevelopmentMetrics for Mofel-Based Systems Development
Metrics for Mofel-Based Systems Development
 
integrating-cognitive-services-into-your-devops-strategy
integrating-cognitive-services-into-your-devops-strategyintegrating-cognitive-services-into-your-devops-strategy
integrating-cognitive-services-into-your-devops-strategy
 
Integrating cognitive services in to your devops strategy
Integrating cognitive services in to your devops strategyIntegrating cognitive services in to your devops strategy
Integrating cognitive services in to your devops strategy
 
Betsol | Machine Learning for IT Project Estimates
Betsol | Machine Learning for IT Project Estimates  Betsol | Machine Learning for IT Project Estimates
Betsol | Machine Learning for IT Project Estimates
 
The future Proof Financial: Fintech
The future Proof Financial: FintechThe future Proof Financial: Fintech
The future Proof Financial: Fintech
 
Doing Analytics Right - Designing and Automating Analytics
Doing Analytics Right - Designing and Automating AnalyticsDoing Analytics Right - Designing and Automating Analytics
Doing Analytics Right - Designing and Automating Analytics
 
Software Project Estimation
Software Project EstimationSoftware Project Estimation
Software Project Estimation
 
Demand Forecasting Case Study ppt - pdf
Demand Forecasting Case Study ppt - pdfDemand Forecasting Case Study ppt - pdf
Demand Forecasting Case Study ppt - pdf
 
Agile Methods: Fact or Fiction
Agile Methods: Fact or FictionAgile Methods: Fact or Fiction
Agile Methods: Fact or Fiction
 
AI improves software testing by Kari Kakkonen at TQS
AI improves software testing by Kari Kakkonen at TQSAI improves software testing by Kari Kakkonen at TQS
AI improves software testing by Kari Kakkonen at TQS
 
Managing Software Project
Managing Software ProjectManaging Software Project
Managing Software Project
 
So Now You’re a UiPath Developer – What’s Next?” What Role do You Play as Dev...
So Now You’re a UiPath Developer – What’s Next?” What Role do You Play as Dev...So Now You’re a UiPath Developer – What’s Next?” What Role do You Play as Dev...
So Now You’re a UiPath Developer – What’s Next?” What Role do You Play as Dev...
 
Mohammed AL Madhani
Mohammed AL MadhaniMohammed AL Madhani
Mohammed AL Madhani
 

More from PMI-Montréal

La mobilisation des équipes de projets pour sortir gagnant de la crise
La mobilisation des équipes de projets pour sortir gagnant de la criseLa mobilisation des équipes de projets pour sortir gagnant de la crise
La mobilisation des équipes de projets pour sortir gagnant de la crise
PMI-Montréal
 

More from PMI-Montréal (20)

La mobilisation des équipes de projets pour sortir gagnant de la crise
La mobilisation des équipes de projets pour sortir gagnant de la criseLa mobilisation des équipes de projets pour sortir gagnant de la crise
La mobilisation des équipes de projets pour sortir gagnant de la crise
 
Adoption du changement : êtes-vous prêts?
Adoption du changement : êtes-vous prêts?Adoption du changement : êtes-vous prêts?
Adoption du changement : êtes-vous prêts?
 
Workshop - Lean change & the gang
Workshop - Lean change & the gangWorkshop - Lean change & the gang
Workshop - Lean change & the gang
 
Mentorat du PMI-Montréal - Séance informative mai 2020
Mentorat du PMI-Montréal - Séance informative mai 2020Mentorat du PMI-Montréal - Séance informative mai 2020
Mentorat du PMI-Montréal - Séance informative mai 2020
 
Désinfection (COVID-19) Ce que vous devez savoir pour un chantier productif
Désinfection (COVID-19) Ce que vous devez savoir pour un chantier productifDésinfection (COVID-19) Ce que vous devez savoir pour un chantier productif
Désinfection (COVID-19) Ce que vous devez savoir pour un chantier productif
 
Leadership responsable : mettez votre masque d’oxygène en premier!
Leadership responsable : mettez votre masque d’oxygène en premier!Leadership responsable : mettez votre masque d’oxygène en premier!
Leadership responsable : mettez votre masque d’oxygène en premier!
 
Delegation Poker - Responsabilisez vos équipes et amenez-les vers une grande ...
Delegation Poker - Responsabilisez vos équipes et amenez-les vers une grande ...Delegation Poker - Responsabilisez vos équipes et amenez-les vers une grande ...
Delegation Poker - Responsabilisez vos équipes et amenez-les vers une grande ...
 
Agile et gestion du changement - Au-delà du Manifeste et de la méthodologie
Agile et gestion du changement -  Au-delà du Manifeste et de la méthodologie Agile et gestion du changement -  Au-delà du Manifeste et de la méthodologie
Agile et gestion du changement - Au-delà du Manifeste et de la méthodologie
 
Agilité comportementale – Comment adapter ses comportements en temps de crise...
Agilité comportementale – Comment adapter ses comportements en temps de crise...Agilité comportementale – Comment adapter ses comportements en temps de crise...
Agilité comportementale – Comment adapter ses comportements en temps de crise...
 
Le Design Thinking : Penser et agir autrement pour trouver des solutions diff...
Le Design Thinking : Penser et agir autrement pour trouver des solutions diff...Le Design Thinking : Penser et agir autrement pour trouver des solutions diff...
Le Design Thinking : Penser et agir autrement pour trouver des solutions diff...
 
COVID-19 et Télétravail - Comment garder votre équipe de projet productive et...
COVID-19 et Télétravail - Comment garder votre équipe de projet productive et...COVID-19 et Télétravail - Comment garder votre équipe de projet productive et...
COVID-19 et Télétravail - Comment garder votre équipe de projet productive et...
 
Matinee PMI-Montréal - Softskills, incontournable pour l'ingénieur en gestion...
Matinee PMI-Montréal - Softskills, incontournable pour l'ingénieur en gestion...Matinee PMI-Montréal - Softskills, incontournable pour l'ingénieur en gestion...
Matinee PMI-Montréal - Softskills, incontournable pour l'ingénieur en gestion...
 
Matinée 11 février 2020 - Priorisation d'un portefeuille de projet
Matinée 11 février 2020 - Priorisation d'un portefeuille de projetMatinée 11 février 2020 - Priorisation d'un portefeuille de projet
Matinée 11 février 2020 - Priorisation d'un portefeuille de projet
 
Comment animer un atelier de gestion de risques?
Comment animer un atelier de gestion de risques?Comment animer un atelier de gestion de risques?
Comment animer un atelier de gestion de risques?
 
Se réapproprier la gestion BIM avec annexes
Se réapproprier la gestion BIM avec annexesSe réapproprier la gestion BIM avec annexes
Se réapproprier la gestion BIM avec annexes
 
MATINÉE - BÂTIR UN PROJET DE VILLE/DESTINATION INTELLIGENTE : ENTRE L'UTOPIE ...
MATINÉE - BÂTIR UN PROJET DE VILLE/DESTINATION INTELLIGENTE : ENTRE L'UTOPIE ...MATINÉE - BÂTIR UN PROJET DE VILLE/DESTINATION INTELLIGENTE : ENTRE L'UTOPIE ...
MATINÉE - BÂTIR UN PROJET DE VILLE/DESTINATION INTELLIGENTE : ENTRE L'UTOPIE ...
 
La gestion de projet dans l'industrie 4.0
La gestion de projet dans l'industrie 4.0La gestion de projet dans l'industrie 4.0
La gestion de projet dans l'industrie 4.0
 
Matinée PMI - De gestionnaire de projet à producteur exécutif!
Matinée PMI - De gestionnaire de projet à producteur exécutif!Matinée PMI - De gestionnaire de projet à producteur exécutif!
Matinée PMI - De gestionnaire de projet à producteur exécutif!
 
PMI-Montréal - protection des données conformité gouvernance 2019 06
PMI-Montréal - protection des données conformité gouvernance 2019 06 PMI-Montréal - protection des données conformité gouvernance 2019 06
PMI-Montréal - protection des données conformité gouvernance 2019 06
 
Symposium 2019 : Quand l'industrie des technologies se mobilise
Symposium 2019 : Quand l'industrie des technologies se mobiliseSymposium 2019 : Quand l'industrie des technologies se mobilise
Symposium 2019 : Quand l'industrie des technologies se mobilise
 

Recently uploaded

The basics of sentences session 3pptx.pptx
The basics of sentences session 3pptx.pptxThe basics of sentences session 3pptx.pptx
The basics of sentences session 3pptx.pptx
heathfieldcps1
 
Vishram Singh - Textbook of Anatomy Upper Limb and Thorax.. Volume 1 (1).pdf
Vishram Singh - Textbook of Anatomy  Upper Limb and Thorax.. Volume 1 (1).pdfVishram Singh - Textbook of Anatomy  Upper Limb and Thorax.. Volume 1 (1).pdf
Vishram Singh - Textbook of Anatomy Upper Limb and Thorax.. Volume 1 (1).pdf
ssuserdda66b
 

Recently uploaded (20)

General Principles of Intellectual Property: Concepts of Intellectual Proper...
General Principles of Intellectual Property: Concepts of Intellectual  Proper...General Principles of Intellectual Property: Concepts of Intellectual  Proper...
General Principles of Intellectual Property: Concepts of Intellectual Proper...
 
Mehran University Newsletter Vol-X, Issue-I, 2024
Mehran University Newsletter Vol-X, Issue-I, 2024Mehran University Newsletter Vol-X, Issue-I, 2024
Mehran University Newsletter Vol-X, Issue-I, 2024
 
Mixin Classes in Odoo 17 How to Extend Models Using Mixin Classes
Mixin Classes in Odoo 17  How to Extend Models Using Mixin ClassesMixin Classes in Odoo 17  How to Extend Models Using Mixin Classes
Mixin Classes in Odoo 17 How to Extend Models Using Mixin Classes
 
TỔNG ÔN TẬP THI VÀO LỚP 10 MÔN TIẾNG ANH NĂM HỌC 2023 - 2024 CÓ ĐÁP ÁN (NGỮ Â...
TỔNG ÔN TẬP THI VÀO LỚP 10 MÔN TIẾNG ANH NĂM HỌC 2023 - 2024 CÓ ĐÁP ÁN (NGỮ Â...TỔNG ÔN TẬP THI VÀO LỚP 10 MÔN TIẾNG ANH NĂM HỌC 2023 - 2024 CÓ ĐÁP ÁN (NGỮ Â...
TỔNG ÔN TẬP THI VÀO LỚP 10 MÔN TIẾNG ANH NĂM HỌC 2023 - 2024 CÓ ĐÁP ÁN (NGỮ Â...
 
Kodo Millet PPT made by Ghanshyam bairwa college of Agriculture kumher bhara...
Kodo Millet  PPT made by Ghanshyam bairwa college of Agriculture kumher bhara...Kodo Millet  PPT made by Ghanshyam bairwa college of Agriculture kumher bhara...
Kodo Millet PPT made by Ghanshyam bairwa college of Agriculture kumher bhara...
 
The basics of sentences session 3pptx.pptx
The basics of sentences session 3pptx.pptxThe basics of sentences session 3pptx.pptx
The basics of sentences session 3pptx.pptx
 
SOC 101 Demonstration of Learning Presentation
SOC 101 Demonstration of Learning PresentationSOC 101 Demonstration of Learning Presentation
SOC 101 Demonstration of Learning Presentation
 
ComPTIA Overview | Comptia Security+ Book SY0-701
ComPTIA Overview | Comptia Security+ Book SY0-701ComPTIA Overview | Comptia Security+ Book SY0-701
ComPTIA Overview | Comptia Security+ Book SY0-701
 
How to Create and Manage Wizard in Odoo 17
How to Create and Manage Wizard in Odoo 17How to Create and Manage Wizard in Odoo 17
How to Create and Manage Wizard in Odoo 17
 
Spatium Project Simulation student brief
Spatium Project Simulation student briefSpatium Project Simulation student brief
Spatium Project Simulation student brief
 
Understanding Accommodations and Modifications
Understanding  Accommodations and ModificationsUnderstanding  Accommodations and Modifications
Understanding Accommodations and Modifications
 
Basic Civil Engineering first year Notes- Chapter 4 Building.pptx
Basic Civil Engineering first year Notes- Chapter 4 Building.pptxBasic Civil Engineering first year Notes- Chapter 4 Building.pptx
Basic Civil Engineering first year Notes- Chapter 4 Building.pptx
 
2024-NATIONAL-LEARNING-CAMP-AND-OTHER.pptx
2024-NATIONAL-LEARNING-CAMP-AND-OTHER.pptx2024-NATIONAL-LEARNING-CAMP-AND-OTHER.pptx
2024-NATIONAL-LEARNING-CAMP-AND-OTHER.pptx
 
Fostering Friendships - Enhancing Social Bonds in the Classroom
Fostering Friendships - Enhancing Social Bonds  in the ClassroomFostering Friendships - Enhancing Social Bonds  in the Classroom
Fostering Friendships - Enhancing Social Bonds in the Classroom
 
Towards a code of practice for AI in AT.pptx
Towards a code of practice for AI in AT.pptxTowards a code of practice for AI in AT.pptx
Towards a code of practice for AI in AT.pptx
 
Google Gemini An AI Revolution in Education.pptx
Google Gemini An AI Revolution in Education.pptxGoogle Gemini An AI Revolution in Education.pptx
Google Gemini An AI Revolution in Education.pptx
 
Vishram Singh - Textbook of Anatomy Upper Limb and Thorax.. Volume 1 (1).pdf
Vishram Singh - Textbook of Anatomy  Upper Limb and Thorax.. Volume 1 (1).pdfVishram Singh - Textbook of Anatomy  Upper Limb and Thorax.. Volume 1 (1).pdf
Vishram Singh - Textbook of Anatomy Upper Limb and Thorax.. Volume 1 (1).pdf
 
Single or Multiple melodic lines structure
Single or Multiple melodic lines structureSingle or Multiple melodic lines structure
Single or Multiple melodic lines structure
 
Unit-IV; Professional Sales Representative (PSR).pptx
Unit-IV; Professional Sales Representative (PSR).pptxUnit-IV; Professional Sales Representative (PSR).pptx
Unit-IV; Professional Sales Representative (PSR).pptx
 
UGC NET Paper 1 Mathematical Reasoning & Aptitude.pdf
UGC NET Paper 1 Mathematical Reasoning & Aptitude.pdfUGC NET Paper 1 Mathematical Reasoning & Aptitude.pdf
UGC NET Paper 1 Mathematical Reasoning & Aptitude.pdf
 

Symposium 2019 : Gestion de projet en Intelligence Artificielle

  • 2. About me 2 Master in Bioinformatics Strasbourg University (France) Ph.D. In Pharmaceutical Science. Strasbourg University (France) Post-Doc at McGill (Computational chemistry) Post-Doc at UdeM (Computational Biology) Senior Data Scientist at Mnubo (IoT company) Nathanael Weill
  • 3. What is AI? Why AI? AI project phases Warnings Optimize the process Outlines 3
  • 4. The theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages. (google dictionary) What is AI? 4 Prediction: The process of filling in missing information. Prediction takes data you have to generate data you don’t have.
  • 5. How does it work? 5 computer Input data Output Function computer Input data Function Output computer New Input data Prediction Function
  • 8.
  • 9. Big Data & Data Science Projects Failure Rate 9 GARTNER ESTIMATED 85% of big data projects fail (2017). The initial estimation was 60% (GARTNER 2016) THROUGH 2020 80% of AI projects will remain alchemy, run by wizards whose talents will not scale in the organization. (GARTNER 2018) THROUGH 2022 20% of analytic insights will deliver business outcomes. (GARTNER 2018) EXECUTIVE SURVEY 77% respondents say that “business adoption” of big data and AI initiatives continues to represent a challenge for their organizations (NEWVANTAGE PARTNERS 2019)
  • 10. A recipe for failure We must define the solution as an entire process. If prediction is the end of the solution, the entire solution might fail because: • The output does not correspond to the operational needs. • The operator will not use it due to complexification of the process. • No one is capable of managing the algorithms if something goes wrong. • … Data Algorithm Prediction
  • 11. Data Algorithm Prediction Judgment Action Feedback Critical! We have to make sure we produce the right information and in the right format to help the person in charge to take action Manager: Person in charge to take action. We need to make sure this person is involved early in the process Design of the solution
  • 12. Identification of the problem to solve Design the appropriate solution Proof of concept Productization Scale the process Reorganize the company 6 Phases 12
  • 14. At Mnubo we designed a 3-5 days workshop with clients to go from the problem identification to the mock up of the solution Performance problem? Scalability issue? How to Consume the predictions? Maintain the solution? What action(s) will be taken? … Ex: 1 prediction per machine? Every hour? 12 hours? Solving the right problem 14
  • 15. A journey as a Data Scientist 1/2 15 Data Scientist: Define the valuable business problem Translate the business problem into a KPI A Key Performance Indicator (KPI) is a measurable value that demonstrates how effectively a company is achieving key business objectives. Organizations use key performance indicators at multiple levels to evaluate their success at reaching targets. Client: « I loose a lot of money when the assembly lines stops ». « I would like to reduce the number of machine failures ». https://www.klipfolio.com/resources/kpi-examples
  • 16. A journey as a Data Scientist 2/2 16 Data Scientist: Define the metric and the definition of success. Next phase: Proof of concept. • explore • Establish a baseline • Iterate!!! Client: A success would be to predict failure 12 hours in advance with an accuracy of 80% According to the final report, I get an answer to: • Is the objectives reasonable? • How should I productize the solution?
  • 17. POC: Critical choice 17 Time Resources Accuracy • Explore • Create a baseline • Iterate Agile
  • 18. Productization phase 18 2 productization models: • Data scientist write specifications and engineers take over and rewrite the code in an other language (java, scala…) • Data scientist with a team of data engineer, dev ops etc… takes the code written and deploy it in the infrastructure Pros and cons…
  • 19. Data Algorithm Prediction Judgment Action Feedback Data Algorithm Prediction Judgment Action Feedback Data Algorithm Prediction Judgment Action Feedback Data Algorithm Prediction Judgment Action Feedback Data Algorithm Prediction Judgment Action Feedback Data Algorithm Prediction Judgment Action Feedback Data Algorithm Prediction Judgment Action Feedback Full solution management: • Configuration • Monitoring • ROI evaluation Scaling of the Solution Avoid silos labyrinthine system
  • 20. Data Algorithm Judgment Action Feedback Data Algorithm Prediction Judgment Action Feedback (Automating)
  • 21. Dev ops: In charge of deploying and maintaining the infrastructure to support the solution Data engineer: in charge of setting the appropriate resource to access the data. Data scientist: in charge of creating the machine learning model (pipeline data to prediction) Roles: development phases 21
  • 22. Operator: In charge of activating/deactivating the algorithms designed for specific predictions/actions => Provide feedback to data scientists Data scientist: Integrate the feedback and update the algorithm (if needed) Dev ops: Maintain the infrastructure Roles: long term 22
  • 24. 24 The Proof of Concept Curse in AI and IoT 80% of companies stop at the POC stage. Laggards & Winners
  • 25. I recommend: To use Agile methodology in all phases of the project Have a clear understanding of the final aim in term of: • The process of development • The perturbation of the company organization Critical role of the project manager 25 Phases: Identification of the problem to solve Design the appropriate solution Proof of concept Productization Scale the process Reorganize the company
  • 26. There is multiple tracks that can be done in parallel: Data acquisition To make sure the data are available in (near-) real time. Creation of the machine learning algorithm Create the appropriate pipeline to train, test and deploy the model(s). Creation of the end point to expose the predictions A dashboard, an app, an alerting system, a reporting system. Monitoring of the pipeline monitor the data acquisition, the performance of the model, the use of the end point… Process to capture the action taken and consolidate a feedback loop Optimize the process 26
  • 27. Hofstadter's law: It always takes longer than you expect, even when you take into account Hofstadter's Law. First AI project is hard, you should start with an easy project • Is there already a system in place to monitor the KPI? • Is the data pipeline already in place? • Is AI a replacement for an existing system? Assess the client maturity is hard especially regarding the company perturbation A good PM is the key to success! Wrap up 27