Tech Talk with Rea and Ashwin Tech Talk

On the season finale, Rea and Ashwin are joined by Sarah Shaik and Nakul Malhotra. The theme of our discussion is 'ethics in technology'

We discuss the example of a smart home speaker (think 'Google Home') and all the ethichal aspects around Google Home's usage and data governance.
Sarah introduces us to the 3 pillars of ethics in technology - 1. rights to data, 2. utilitarianism and 3. Common good while Nakul talks about how there is a need to be cautiously optimistic while developing and adopting technology.
We also talk about the growing influence of the big techs and their responsibility to be ethical while scaling. 

Our guests, Sarah and Nakul work as software engineers/developers in California and are quite passionate about the ethical side of tech.

Their LinkedIn's-

Sarah Shaik- https://www.linkedin.com/in/sarah-shaik/

Nakul Malhotra- https://www.linkedin.com/in/nakulmalhotra/

Links- 

Google Assistant calling a restaurant for a reservation
https://www.youtube.com/watch?v=-RHG5DFAjp8

On the 7th episode, we recap Neural Networks and introduce Deep Neural Networks / Deep Learning. We discuss the advantages of Deep Learning , its real-world applications, and how they are different from neural networks. Deep Learning is compute-intensive and when used well, can save thousands of hours while trying to make predictions.

Show notes-
Edu material -
Neural Networks in 5 minutes - https://www.youtube.com/watch?v=bfmFfD2RIcg

How Deep Neural Networks Work - https://www.youtube.com/watch?v=ILsA4nyG7I0

Relevant timestamps-
00.15-2.30 - Recap of Neural Networks
2.30-4.30 - Introduction to topic and business application
4.30- 7.30 - Introduction to deep learning (with retail example)
7.30- Behind the scenes-Neural Networks
10.00- 10.45- Difference between neural networks and deep learning
10.45- 14.00 - How Deep Learning Is self-teaching
14.00- 15.00 Deep Learning and its resource instensive nature
15.00 - 17.15- Benefits of Cloud in enabling deep learning
17.15-19.30 - Summary
19.30-20.00- Special Announcement and wrap up  

On episode 6, we discuss some common day-to-day examples of where AI and ML are being used. For example - Spotify's Discovery feature, Youtube Video Recommendations, Facial Recognition.

We deep dive into the example of how AI and ML tools can help with predicting the value of a property. We understand why and how more data (and inputs) increases the accuracy of whatever we are trying to predict (in this case, the property value).

We also throw in some mild regression and stats references to sound smart but the episode remains simple for the most part to introduce the concept of neural networks.

We discuss the use of these neural networks (a series of algorithms) that enable the AI and ML predictions. We discuss their application in retail (Online shopping like Instacart and Sephora) and how these tools help for better customer repeat purchases and retention. Other industries we touch are real estate as well as raiways.

"The more Instacart you use, the better Instacart gets at predicting the items you would want to shop for."

Edu material-

A relevant Youtube video that summarizes AI/ML - https://www.youtube.com/watch?v=mJeNghZXtMo

Timestamps -

0.30' - 1.30' - Topic Introduction
1.30' - 4.00'- Various day-to-day examples
4.00' - 9.00'- Application in Real Estate
10.00' - 13.00' Application in Retail (Instacart and Sephora)
13.00' - 14.00'- Privacy Concerns around data collection
 14.00'- 15 Online Grocery Store Ocado using Dynamic Pricing to decide prices
15.15'- 16.00' How these predictions get better (investment to improve algorithm accuracy by reducing margin of error)
16.00 - 17.30' Application in Railways
17.30' -  19.10' - Summary and wrap up.

On episode 5, we do a brief recap of episodes 1 to 4 and solidify the learnings by discussing the NoSQL Document Database and its application in a real-life marketing scenario.

Have you ever gotten targeted with images of a product on social media only minutes after you mentioned the product to a friend or browsed it on your phone? Listen more to find out how this works!

Relevant timestamps-
0'-1.20'- Recap of 1-4 episodes
For the rest of the episode, we discuss a real-life scenario of a consumer being shown images of products on social media that he/she may have browsed only minutes ago. We understand how a NoSQL Document database enables this 'smart marketing'.

We chat about ETL (Export, Transfer and Load) Constraints, Formatting Constraints and other constraints that make SQL databases less favorable than NoSQL Databases for the above mentioned application.

Reference Video to understand NoSQL Document Database- https://www.youtube.com/watch?v=nigPkP6QeTk

On episode 4, we discuss the types of Relational (SQL) Databases provided by different Cloud Providers and their unique benefits. We break down the terms 'cloud native' and 'cloud capable' services. We discuss ROI, one of the key metrics that helps a company in making a decision about moving to the cloud or not. We also chat about data warehousing and reiterate the value of the cloud.

Relevant links-
Amazon Docs - https://docs.aws.amazon.com/
Amazon ROI Calculator- https://aws.amazon.com/tco-calculator/

Microsoft Docs- https://docs.microsoft.com/en-us/azure/?product=featured
Microsoft Calculator-  https://azure.microsoft.com/en-us/pricing/tco/calculator/

Relevant timestamps:
0-1' Recap
1'-2:30'- Inroducing topic
2.30'-3.20'- Cloud Native Services
3.30-4.15 Cloud Capable Services (Eg- Mongo DB)
4.15'- 7.15'- Integration of cloud native and cloud capable services
7.15'- 8.30 Amazon's Relational Database Offering (RDS and Redshift among others)
8.30' - 9.20' - Data Warehousing
9.20' - 12.20' - Microsoft's Relational Database Offering 
12.20'-13.30'- Business Case for databases
13.30-14'- Google's offerings
14-17.30' - ROI 
17.30' - 20' - Long term benefits of Cloud
20'-22'- Recap

On the third episode of the show, we talk about the different layers of an application and specifically focus on databases. We introduce and breakdown 'SQL' and 'NoSQL' databases. We also talk about their use-cases and benefits while introducing examples like Netflix. We chat about ACID compliance and wrap the episode up, with a quick summary.

Recommended reading- A link that summarises SQL v/s NoSQL pretty well- https://www.thorntech.com/2019/03/sql-vs-nosql/

Relevant timestamps: 
0:30-2:00-Recap of episode 2
2:00-3:15 Different kinds of layers in an application (Presentation, Application and Database)
3:15-4:00 - Introduction to Databases
4:00- 6:00- SQL
6:00-11:15- NoSQL and types of NoSQL Databases (Netflix Example)
11:15-13:00- Document Databases
13:00- 15:00 Wide column databases
15:00-16:00- Graph Databases
16:00-16:40 - How to decide between SQL and NoSQL Databases
16:40-22:00- ACID Compliance 
22:00-24:30- Wrap up and Summary