1.
IT Industry
How to Getting Into IT With Zero Experience?
Eng. Rafie Tarabay
eng_rafie@mans.edu.eg

2.
• Head of Students' scientific society 1997 to
2000
• Head of Programmers Syndicate DK Branch
from 2011 till now.
• Work on GULF 4 years, and on IBM 10 years
• Work as a Senior Engineer on Dubai
(Mohammed Bin Rashid Al Maktoum Knowledge
Foundation) for 2 years
• Develop many enterprise applications that
server Egyptian Government and more that 100
Organization/Universities on MEA
• Application Development Consultant
21+ years Full stack developer, Certified ITIL
• Expert in Bigdata, Machine Learning, Block Chain, and
IOT
• Have 4 Mastery Award from IBM
1) IBM Artificial Intelligence Analyst Mastery Award
2) IBM Predictive Analytics Mastery Award
3) IBM IOT Mastery Award
4) IBM Blockchain Mastery Award

3.
Where to go?
1. Help desk position: ICDL, ITIL
2. Tester: ISTQB
3. SystemAdministrator:
1. MCSE
2. CCNA
3. Linux
4. DevOPs (Docker , Kubernetes)
5. Cloud
4. Application Developer

4.
Introduction to Development Path
About System Analysis Steps
PMP vs Agile (PM vs PO, and Scrum Master)
Career path: Developer, Senior Developer ,Team Lead, System Analyst,
System Architecture, Project Manager
Mobile
Flutter, React-Native
Web-Technology
• Backend: PHP, Java, NodeJs, Python, ASP.net core 6.0
• Frontend: Angular, VueJS, React + Bootstrap 5.0+HTML, CSS, JavaScript
Database
SQL Server, DB2 Oracle
MySQL/ MariaDB, MongoDB
Advanced Topics
MQ/Kafka , Solr 8, Blockchain, Big Data/Hadoop, Redis

5.
In Software Development
1. Start small then extend
2. Change one thing at a time
3. Add error handling early
4. Test the part before whole
5. Before change, first understand the existing code
6. There will always be bugs
7. Ask, ask, and ask
8. Keep learning

6.
Open Source
ERP: Odoo, ERP Next
eCommerce: NopCommerce, GrandNode
Portals: WordPress, Joomla

7.
Cloud
1. MS Azure
2. Amazon AWS
3. Google Cloud
4. IBM Cloud
5. Digital Ocean

8.
AI industry
1. Software Engineer
2. Data Engineer
3. Data Science
4. Data Analysis

9.
Machine learning techniques
Machine learning mainly has three types of learning techniques:
• Supervised learning
• Unsupervised learning
• Reinforcement learning

10.
Accuracy vs Precision
 Accuracy is how close a measured value is to the actual (true) value.
 Precision is how close the measured values are to each other.

11.
Machine Learning tasks categories
1. Classification
2. Regression
3. Clustering
4. Anomaly detection
5. Association
6. Recommendation
7. Dimensionality reduction
8. Computer Vision
9. Text Analytics

12.
Common modeling techniques
Supervised Learning
• describe and distinguishes classes for future prediction (on new data) based on training data
• Classification & Prediction
• Common Methods: Decision Trees, Regression, Nearest Neighbours, Neural Networks
Unsupervised Learning
• Analyses data where labels are unknown to create groups/classes for objects that are similar to each other
within the group but dissimilar to objects in other clusters
• Cluster
• Common Methods: K-means, Hierarchical, Two-Step
Association
• Analyzing data to detect items that occur together
• Association
• Common Methods Apriori, CARMA

13.
Classification [supervised learning]
• Classification is concerned with building models that separate data into distinct classes. These models are
built by inputting a set of training data for which the classes are pre-labelled in order for the algorithm to
learn from. The model is then used by inputting a different dataset for which the classes are withheld,
allowing the model to predict their class membership based on what it has learned from the training set.
Binary classification examples (divide data to two options only)
• spam filtering is a classification task
• Tumor diagnosis can be treated as a classification problem.
• determining credit risk using personal information such as income, outstanding debt
• use ROC Curves (Receiver operator characteristics) to summarize & present performance, models distinguish between
false & true positives
Multi-class classification examples
• handwritten recognition each character is a multi-class classification problem
• image recognition is a multi-class classification task
• Xbox Kinect360, which infers body parts and position

14.
Regression [ supervised learning]
• The goal is to predict a numerical label for an unlabeled observation
Regression algorithms: Linear regression, Decision trees
Examples
• home valuation
• Asset trading, and forecasting
• Sales or inventory forecasting

15.
Anomaly Detection [ supervised learning]
the goal is to find outliers, noise, deviations in a dataset
Anomaly detection applications
• In manufacturing, it is used for automatically finding defective products.
• In data centers, it is used for detecting bad systems.
• Websites use it for fraud detection.
• Detecting security attacks. Network traffic associated with a security attack is unlike normal
network traffic. Similarly, hacker activity on a machine will be different from a normal user
activity.

16.
Clustering [unsupervised learning]
• The aim is to split a dataset into a specified number of clusters or segments.
• Elements in the same cluster are more similar to each other than to those in other clusters.
• Elbow diagram we use to find the best no of clusters
Clustering algorithms
1) k-means
• The number of clusters (k) must be given explicitly.
• Identify the best k cluster centers in an iterative manner
• Clusters are assumed to be spherical.
2) OPTICS /DBSCAN
• it is a density-based clustering algorithm. represents clusters by its nature
Example
• creating customer segments, which can be targeted with different marketing programs

17.
Association [unsupervised learning]
• Association is most easily explained by introducing market basket analysis, a typical task for
which it is well-known.
• attempts to identify associations between the various items that have been chosen by a
particular shopper and placed in their market basket and assigns support and confidence
measures for comparison.
• The value of this lies in cross-marketing and customer behavior analysis.

18.
Dimensionality Reduction [unsupervised learning]
• The goal in dimensionality reduction is to reduce the number of features in a dataset without
significantly impacting the predictive performance of a model and this will reduce the
computational complexity and cost of machine learning.

19.
Recommendation [Reinforcement]
• The goal of a recommendation system is to recommend a product to a user based on past
behavior to determine user preferences.
• Unlike Association, Recommendation focus on user behavior and suggest according to this user
behavior.
• It is reinforcement because we not sure of result until user choose one of our recommendation, if
not, then our recommendation was not correct.

20.
Traditional ML Algorithms
• can work on low-end machines
• break the problem down into different parts, solve
them individually and combine them to get the
result.
• takes much less time to train, ranging from a few
seconds to a few hours.
• gives us the selection rules, so it is easy to interpret,
safe to use in industry for interpretability.
Deep learning algorithms
• need a large amount of data to work perfect
• heavily depend on GPUs
• solve the problem end-to-end.
• takes a long time to train. (like two weeks)
• It is excellent and is near human performance, but
no guarantee to be always like this!, because most
of operations are hidden and we can’t review
selection logic!

21.
Deep Learning Birthdate
ImageNet challenge: It is Olympics of computer vision!, Every
year, researchers attempt to classify images into one of 200 possible
classes given a training dataset of approximately 450,000 images.
The goal of the competition is to push the state of the art in
computer vision to rival the accuracy of human vision itself
(approximately 95– 96%).
In 2012, Alex Krizhevsky at the University of Toronto did the
unthinkable. Pioneering a deep learning architecture known as a
convolutional neural network for the first time on a challenge of
this size and complexity, he blew the competition out of the water.
The runner up in the competition scored a commendable 26.1%
error rate. But AlexNet, over the course of just a few months of
work, completely crushed 50 years of traditional computer vision
research with an error rate of approximately 16%

22.
Deep learning CNN/RNN
CNN is a feed forward neural network that is generally used for Image recognition and object classification.
CNN considers only the current input
while RNN considers the current input and also the previously received inputs. It can memorize previous inputs due to its
internal memory.
Use CNNs For: Image data
Use RNNs For: Text data, Speech data

23.
Deep Learning Example
GFP-GAN is a new free AI model for photo restoration.

24.
Deep Learning Example
A new GAN algorithm to composite the 'real' images, MixNMatch, from University of California.
MixNMatch learns the background, object pose, shape, and texture from real images with minimal
supervision, to mix and generated realistic images.
Read more here, MixNMatch:
https://arxiv.org/abs/1911.11758
GAN framework:
https://arxiv.org/abs/1811.11155

25.
Online Demo
How to build a simple CURD
How to publish to AWS

26.
Questions & answers
Invite questions from the audience.

27.
IT Industry
How to Getting Into IT With Zero Experience?
Eng. Rafie Tarabay
eng_rafie@mans.edu.eg