Iceberg, Right Ahead!!!
being Future Ready
Gopi Krishna Nuti
Lead Data Scientist
Autodesk Construction Solutions
gopi.nuti@autodesk.com
Alumnus, AU College of Engineering 1998-2002
ngopikrishna.public@gmail.com
+91-9036005121
Do you
recognize
these
quotes?
“Houston, we have a
problem.”
"Bond. James Bond."
"I'll be back."
“Iceberg!!! Right
Ahead!!!”
Iceberg! Right Ahead!
(Image: © Shutterstock)
You cannot become what you need to be
By remaining what you are.
–-- Max de Pree
Expectations
T-shaped Skills
How to identify
Emerging Technologies?
Gartner© Hype cycle
Image courtesy https://blogs.gartner.com
Hype Cycle
Risk is high during the
Peak & Trough stages.
Risk is low during Slope
and Very Low during
the plateau. BUT there
is no more gold mine.
Where can we learn these?
Internet, of
course!!!
Free Resources
EdX NPTEL Tutorials Point
Coursera
Lots and lots of
MOOCs.
Training
material is
growing
exponentially
Soft skills
Reading makes a full man; Conference a ready
man; and writing an exact man.
--- Sir Francis Bacon
Full – Does not mean “complete”. It is the opposite of “Empty”
Conference – means Conversation
Where can I gain these?
Internet, of course???
NO. of course,
not!!!
How to gain these
Seminars
Soft
skills
Why do I need these?
Simply put, to gain a USP.
A reason for the recruiter to choose
you instead of your classmate.
Why do I need these?
• Put yourselves in the shoes of the recruiter and look at
your own profile objectively.
• No emotional statements
• No hypocrisy
• Aim for the stars. You might land on the moon.
Easy-to-learn-yet-high-return topics
Data Science, AI/ML
Cloud Computing
Edge Analytics
GPU accelerators – Both Programming and Hardware
Augmented Reality/ Virtual Reality
Robotic Process Automation
Robotics
Arduino, Raspberry Pi, NVIDIA Jetson
Micro controllers
Internet of Things
Communication Protocols
3G, 4G, 5G, Bluetooth
Mobile app development
Web development
Web
Resources
Wikipedia
www.medium.com
www.towardsdatascience.com
http://electrical-engineering-portal.com/
http://electronics.wisc-online.com/
http://www.electrical4u.com/
http://www.allaboutcircuits.com/
http://vlab.co.in/
https://ocw.mit.edu/index.htm
http://www.electronicsweekly.com/
Specific into
Machine
Learning
Gartner’s AI Hype Cycle for 2020
Growth areas
Newer
Domains
 Traditional - Market Research,
Finance, Engineering,
Education, Medicine,
Astronomy
 New Domains
 Robotic Process Automation
 Warfare and Defence - Drones,
Quadcopters
 Cyber security – Filtering
content, identifying attack
vectors, Threat exposure,
incident response etc.
 Media and Entertainment -
Deep fakes, GANs, Deep
Nostalgia
 Advertising and Marketing-
Mass Customization
 Hybrid Workforce
 Sub disciplines
 ML Ops
AI Market
places
 A multi-sided-platform where
AI providers and sellers can
exchange services. NB:
Seller is not the Cloud
provider!!!
 Clever blend of PaaS & SaaS.
Can be thought of as
Algorithm as a Service
 Build an AI algorithm for a
specific use case. Sell it for a
license on top of cloud
Services.
 Examples
 AWS Marketplace
 Azure Marketplace
AI Marketplace examples
Demonstration of the Mask Detector for Epidemiological Safety by Vitech
Lab. Picture courtesy Sandro Luck
Demonstration of the Vehicle Damage Inspection developed by Persistent
on Amazon Market Place
Anything is game
AI models
Datasets
Data pipeline
management
ML Ops
AI Marketplace examples
GluonCV YOLOv3 Object Detector
By: Amazon Web Services YOLOv3 is a powerful network for fast and
accurate object detection, powered by GluonCV.
Emotion Analysis
API
Sold by:Twinword
Inc.
Deep Vision API
Sold by:Deep Vision AI, Inc
Deep Vision API is a computer vision platform allowing you to
easily integrate AI-based technology into your products, services
and applications. You can automatically understand and analyze
images and videos.Pay directly from your AWS account, quick and
seamless integration with your AWS workflow.
AI Services Commoditization
Image courtesy : Gartner.com
ML algorithms are run on the hardware device/sensor itself.
This is in direct contrast to cloud/server based processing of data.
Processing of data happens close to the user.
Can be on IoT senser or dedicated Edge Server
Benefits
Latency time is reduced my many orders of magnitude
CapEx and OpEx costs are reduced significantly
Increased data privacy and security
Examples
Alexa, Google Assistant,
Self driving cars
Edge AI Camera from Avinton, VIA Mobile360 Dash Cam
Edge AI
Originally thought of as unrelated/competing technologies
Now complementing one another
Robotic Process Automation uses specially developed programs to
automate repeatable business processes. There is no learning. Repeats
the same set of actions every time.
Great for automating simple tasks.
Typically, suitable for structured data
For complex tasks, AI comes handy because
Learns from data/past
Handles unstructured data
Robotic Process Automation and
AI
Specialized hardware to accelerate the processing of AI algorithms
particularly, Deep Neural Networks
GPUs
Nvidia, Intel
 Vision Processing Units
Intel Neural Computing Stick, Qualcomm Snapdragon
 Tensor Processing Units
 FPGAs and ASICs
 Special Purpose programming languages
 CUDA, OpenCL
 OpenVino Kit from Intel, SNPE SDK from Qualcomm
GPU Accelerators
Good Luck

Emerging Technology trends and employability skills

  • 1.
    Iceberg, Right Ahead!!! beingFuture Ready Gopi Krishna Nuti Lead Data Scientist Autodesk Construction Solutions gopi.nuti@autodesk.com Alumnus, AU College of Engineering 1998-2002 ngopikrishna.public@gmail.com +91-9036005121
  • 2.
    Do you recognize these quotes? “Houston, wehave a problem.” "Bond. James Bond." "I'll be back." “Iceberg!!! Right Ahead!!!”
  • 3.
  • 4.
    You cannot becomewhat you need to be By remaining what you are. –-- Max de Pree
  • 5.
  • 6.
  • 7.
  • 8.
  • 9.
  • 12.
    Hype Cycle Risk ishigh during the Peak & Trough stages. Risk is low during Slope and Very Low during the plateau. BUT there is no more gold mine.
  • 13.
    Where can welearn these? Internet, of course!!!
  • 14.
    Free Resources EdX NPTELTutorials Point Coursera Lots and lots of MOOCs. Training material is growing exponentially
  • 15.
    Soft skills Reading makesa full man; Conference a ready man; and writing an exact man. --- Sir Francis Bacon Full – Does not mean “complete”. It is the opposite of “Empty” Conference – means Conversation
  • 16.
    Where can Igain these? Internet, of course??? NO. of course, not!!!
  • 17.
    How to gainthese Seminars Soft skills
  • 18.
    Why do Ineed these? Simply put, to gain a USP. A reason for the recruiter to choose you instead of your classmate.
  • 19.
    Why do Ineed these? • Put yourselves in the shoes of the recruiter and look at your own profile objectively. • No emotional statements • No hypocrisy • Aim for the stars. You might land on the moon.
  • 21.
    Easy-to-learn-yet-high-return topics Data Science,AI/ML Cloud Computing Edge Analytics GPU accelerators – Both Programming and Hardware Augmented Reality/ Virtual Reality Robotic Process Automation Robotics Arduino, Raspberry Pi, NVIDIA Jetson Micro controllers Internet of Things Communication Protocols 3G, 4G, 5G, Bluetooth Mobile app development Web development
  • 22.
  • 23.
  • 24.
    Gartner’s AI HypeCycle for 2020
  • 25.
  • 26.
    Newer Domains  Traditional -Market Research, Finance, Engineering, Education, Medicine, Astronomy  New Domains  Robotic Process Automation  Warfare and Defence - Drones, Quadcopters  Cyber security – Filtering content, identifying attack vectors, Threat exposure, incident response etc.  Media and Entertainment - Deep fakes, GANs, Deep Nostalgia  Advertising and Marketing- Mass Customization  Hybrid Workforce  Sub disciplines  ML Ops
  • 27.
    AI Market places  Amulti-sided-platform where AI providers and sellers can exchange services. NB: Seller is not the Cloud provider!!!  Clever blend of PaaS & SaaS. Can be thought of as Algorithm as a Service  Build an AI algorithm for a specific use case. Sell it for a license on top of cloud Services.  Examples  AWS Marketplace  Azure Marketplace
  • 28.
    AI Marketplace examples Demonstrationof the Mask Detector for Epidemiological Safety by Vitech Lab. Picture courtesy Sandro Luck Demonstration of the Vehicle Damage Inspection developed by Persistent on Amazon Market Place
  • 29.
    Anything is game AImodels Datasets Data pipeline management ML Ops AI Marketplace examples GluonCV YOLOv3 Object Detector By: Amazon Web Services YOLOv3 is a powerful network for fast and accurate object detection, powered by GluonCV. Emotion Analysis API Sold by:Twinword Inc. Deep Vision API Sold by:Deep Vision AI, Inc Deep Vision API is a computer vision platform allowing you to easily integrate AI-based technology into your products, services and applications. You can automatically understand and analyze images and videos.Pay directly from your AWS account, quick and seamless integration with your AWS workflow.
  • 30.
    AI Services Commoditization Imagecourtesy : Gartner.com
  • 31.
    ML algorithms arerun on the hardware device/sensor itself. This is in direct contrast to cloud/server based processing of data. Processing of data happens close to the user. Can be on IoT senser or dedicated Edge Server Benefits Latency time is reduced my many orders of magnitude CapEx and OpEx costs are reduced significantly Increased data privacy and security Examples Alexa, Google Assistant, Self driving cars Edge AI Camera from Avinton, VIA Mobile360 Dash Cam Edge AI
  • 32.
    Originally thought ofas unrelated/competing technologies Now complementing one another Robotic Process Automation uses specially developed programs to automate repeatable business processes. There is no learning. Repeats the same set of actions every time. Great for automating simple tasks. Typically, suitable for structured data For complex tasks, AI comes handy because Learns from data/past Handles unstructured data Robotic Process Automation and AI
  • 33.
    Specialized hardware toaccelerate the processing of AI algorithms particularly, Deep Neural Networks GPUs Nvidia, Intel  Vision Processing Units Intel Neural Computing Stick, Qualcomm Snapdragon  Tensor Processing Units  FPGAs and ASICs  Special Purpose programming languages  CUDA, OpenCL  OpenVino Kit from Intel, SNPE SDK from Qualcomm GPU Accelerators
  • 34.