Technical Interview Questions

This one is reviewed but I need to delete its copy from hubpages or somewhere

NoSQL Data models:

key-value

Aggregate model.
key or id is used to get the data.
Lookup is based on the key.

document

Aggregate model.
key or id used to get the data.
Lookup is based on the fields in the aggregate in form of query. Parts of the aggregate are retrieved rather than whole. DB will create indexes on the aggregate.

People can put ids for lookup.
Mostly the lookup will be on something else and not id.
Partial lookups are possible.

column family

Data is organized into columns.
Useful in case of reads.
Imposes structure on the aggregate.

graph

Data distribution Model:

Sharding: Distributes different data across different servers.
Replication: Copies data across multiple servers. Configurations include:

Master-Slave:

Master handles writes and sync with Slave
Slave handles reads
Reduces the chance of update conflicts.

Peer-to-peer:

Any handle/system will be able to handle the writes. They will sync with each other.
Avoids single point of failure but will need to take care of conflicts.

DB scaling Tips & Tricks:

Don't use shared databases for different applications. Instead use single database for single applications. Then get the data from different applications into Hadoop file system(shared Database).
Vertical Scaling: It means adding disk space, storage, memory.
Horizontal Scaling: It means adding more machines/clusters.
Relational databases can't work on clusters because they communicate to the single disk space. So by desing they will follow vertical scaling.

For different sets of data, rdbms runs on different servers(effectively sharding the data).
Application layer takes care of which data is stored where.
RDBMS are usually single server and negotiation of contracts get tedious.
They are aggregate ignorant.

Polygot persistence: It means using different data stores in different circumstances.
NoSQL is used as application DB and not integration DB.
Aggregate orientation ensures that minimum number of nodes are queried to get the data.

It helps to aggregate data together to ensure that similar data lies on single node so querying is easy and optimized.
Its central to cluster.
DB will ensure data is stored together on nodes as aggregates.
Manipulation of multiple aggregate need to happen at the application level.

Important Metrics for Product Managers

There are 3 broad categories:

User engagement

User Engagement metrics: Use cohort analysis over a period of time. This will help understand how the behavior has changed overtime.

Number of sessions/user.

How often the users log in or open the app.

Session duration for a cohort(group of users performing a specific function) over time:

How much time they spent interacting with your product.
Compare average session duration between churned and retained users.

Number of key user actions/session

Select important user actions(like click on something) then trace them over time and for different groups of people.

Business metrics

Customer Lifetime Value - LTV

Amount of revenue generated by different customers.
Don't calculate Customer Lifetime Revenue.

Deduct operating and development costs.

Customer Acquisition Cost - CAC

Acquisition marketing costs/number of paying customers acquired over period of time.
Acquisition costs = all operating and dev costs.
For healthy SaaS startups LTV/CAC >=3

Average Revenue/Account - ARPA

Amount of revenue per account per month.
If this metric is stable then team is on right track to provide value to customers that keeps them on the platform.
If you have better support and branding then you can consider increasing ARPA without changing pricing or product.

Monthly Recurring Revenue - MRR

Company's or product revenue/month.

Logo churn & Revenue churn rates:

Percentage of paying customers/account that are lost/month.
Calculate as: cohort and aggregated.
Revenue churn: Revenue the company loses/month due to churns or downgrades. This is important.

Retention rate:

Customer Service:

Number of incoming support tickets: Smaller the number better the state of the platform or software.
Net Promoter Score

Source: https://productcoalition.com/critical-metrics-every-product-manager-must-track-c5f1e46e3423

Some math basics for metric calculation:

Mean = usual average of all the numbers

Ex: Mean of 1,2,3,4,5 = (1+2+3+4+5)/5 = 3

Median is the middle value of all the sorted numbers from the input sources:

Ex: 1,2,3,4,5: Median =3
Ex: 1,2,3,4: Median = (2+3)/2 = 2.5

Mode is number that is often repeated:

Ex: 1,3,3,3,4: Mode = 3
Ex: 1,3,4,5: Mode = none since none of the numbers are repeated.

Range: Difference of smallest and largest number

1,2,3,4,5 = 5-1 = 4

List of Product manager questions

One executive says that Feature A is more important and another executive says Feature B is more important. How do you choose which one to implement?
Talk about how you overcame product failures/challenges or poor feedback.
What’s the best way to work with executives?
How do you think product managers interact with engineers?
Tell me about a time when you had to build or motivate a team.
What is a product you currently use everyday, why and how would you improve it?
How do you know if a product is well designed?
Suggest a new feature for Amazon. What metrics would you use measure it’s success?
What metrics do you think would be important for us to track?
How many iPads are sold in the U.S. every year?
How do you think we came up with the price for _____ product?
How many windows are in New York City?
How many people are currently online in Europe?

Distributed Datastore Basics

Basics:

Distributed DB is divided into different nodes & Data is spread across the nodes.
There is no Master/Slave config in distributed DB.
Every machine is given the responsibility of saving part of the data based on some logic.
If some node goes down its fine since some of its data is captured into some other node.
Primary responsibility of a node is to store all the data. Secondary responsibility of another node(backup) will be to store part of that data. So a primary node will store all the data and several secondary nodes will store parts of the data. In case of loss, data will be retrieved from several secondary nodes.
Consistent hashing concept is used to distribute the data across several nodes.

It will determine the hash of data and spread it across different partition keys.

Primary datastore will take care of hashing the partition key and finding the respective node for storing it.
In No-SQL datastores like Cassandra there is a replication factor, which is usually a number. Let's say replication factor is 3 then the data will be stored onto 3 different nodes.

Let's say the backup strategy is to store the data into the next available node. So in that case the data will be stored in the computed node, then next available node, then next available.
So its a combination of consistent hashing and backing up the data into different nodes.

To handle more load extra nodes will be added. This will automatically assign the responsibility of the distributing the data across new and existing nodes using Gossip protocol.

All nodes talk to each other so they everything about each other.
If a node goes down it will not hear or talk anything. When other will communicate to it there will be no response. They'll understand that the node is down so they will share the responsibility amongst themselves automatically. Another node will become primary node to hold the data of that range let's say 10-20k.

Similar strategies are implemented in Redis, HDFS, Cassandra,

HDFS has master node/name node and data node. Data resides in the data node.
In such cases ZooKeeper can take care of assigning responsibilities to slave nodes if master goes down.

Commodity computers are like normal desktop that we use at home but with extra capabilities.

Why you need to scale?

Performance is degraded as customers are complaining that requests are taking more time.

Vertical Scaling:

Upgrade the server to handle volume of requests easily. Upgrading single server's capability much more.

Horizontal Scaling:
1. Add more machines to the cluster.
2. Cheaper than vertical scaling.
3. If service is not heavily used then more machines can be removed from the cluster. Service is not impacted since you are adding more machines to direct the traffic and removing the machines so requests are served by some or the other machine at any given point.
4. Moreover, we can store the machines in different continents.
Conclusion:
1. Vertical scaling is cost efficient in short term as you are buy less H/W. Horizontal scaling is cost efficient in long run.
2. Vertical scaling is not fault tolerant as its only one machine.
Advantages of distributed systems:
1. Fault tolerance:
2. Low Latency:

How to answer behavioral questions?

Why should we hire you?

To answer this questions, specify why are you a good fit to the role and the company.
If you doing some volunteer activities and if it matches the responsibilities of the company then you can specify that. For example, I regularly volunteer at 'X' organization. It requires me to work on some many things at the same time like marketing, sales, event planning, people management, resource management. I enjoy doing all of these activities at once. This role requires me to handle people & marketing hence based on my volunteer experience, I believe I am a good fit!

Why are you leaving your current job?

To answer this question, don't specify that you are bored at your current organization. Instead specify something like this: After being there for 3 years, learning has slowed down. Company isn't growing that much and it is impacting my growth.
If you are looking for stability then mention that. If you are looking for a role with specific skills then mention that.
Always mention something along the line of what you want to do and how the current organization is not offering it!

What do you like to do in your spare time?

To answer such questions try to come across as someone who shows initiative & who is willing to try out multiple things.
These days the roles are not one-dimensional. Instead they are multi-dimensional and expect people to have experience in wide variety of areas. So mention something like:

Working on a side project
Volunteering
Online learning
Hobbies
Working out, cooking, etc

Where do you want to see yourself in next five years?
Interviewers ask this to:

Understand if you really want this job.
See if you have any plan for the future!
Test your ambition.
Ensure the company can give you what you want.

Answers can include that you want to be a Lead person of a growing department or organization within the company. You might want to be in a specific domain. So to answer this question always have a 2,3,5 year plan ready. If you are interviewing for a higher roles then have 10 year plan ready!

Product Manager Q&A

What is the best way to interface with customers?

Face to face meetings

I believe face to face meetings are the best if you have the luxury to meet your customers. During your f2f meetings, you can see how your customers are using your product and you will be able to gain some valuable insights.
For ex: If you have a business product then ask for customers for the flow diagrams that include your product in the flow. Ask them about any client proposals that they are working on and have worked on. Study those as nuggets of information can be found there. As a data PM, it is always useful to read these plans to understand how and where your products are used.

Phone calls

If f2f meetings are not possible then go with phone call. Try to schedule a regular phone call with your customers to understand the pain points.
Ask them directly: Is there something you would like us to improve for you. What works best for you? What needs to be improved? Most often they have their answers ready.

Emails

In come cases, just the emailing works. If this is the case then make them a priority and respond them within reasonable times. Sooner the better.! This shows them they you care about delivering a quality product.
If they are requesting a feature from you and you know its going to take time, try to be open about the timelines. Ask them if there is anything they would like you to do while the much needed feature is in process. This softens the customers a bit and they are open for negotiation.

---------------------------------------------------------------------------------------------------------------------

What's the best way to work with the executives?

Executives are busy people and they appreciate brevity. What is important to you might not be important to them. So we always have to be mindful of their time.
They care a lot about strategy than tactics. They care about business and product goals more than day to day execution.
Any suggestions that move the needle in direction of business goals will gain their attention. So to make any interaction valuable with them, focus on strategy and tell them on how you can help them better achieve the goals.
While talking to them its important to talk about revenue generation, strategy, important KPIs that relate back to strategy, product metrics that are helping achieve the KPIs.
While working on a problem, they are looking for solutions and which solution will work best for them. They usually care less about day to day execution but its best to have that information handy when asked!
Don't provide wrong answers to them. If you don't know, tell them you will get back to them after looking into it.

To summarize while working with executives:

Talk about:

Strategy
KPIs
Business and Product Goals
Solutions to existing problems.

Avoid talking about:

Tactics
Only problems
Wrong answers

---------------------------------------------------------------------------------------------------

What's your approach to hiring?

Hiring the best is always the motto of any organization but other dimension to hiring is 'How well this person will solve the business problem at hand'. Here are few common dimensions to consider while hiring:

Business goals

Let's say my team is really swamped and they could use some help then I will hire someone with specific skills set. I will work with the team manager to identify a specific skill-set that will benefit the team. And if I know exactly what projects we have to deliver and if we get a person with the exact skill set then I will go ahead with that person.
However if I am hiring for an long term business goals then I will hire someone with determination and eagerness to learn. Someone who might not have demonstrated bigger professional achievements but is willing to explore and take on challenges. I will look for someone who has done some projects on the side that explains their motivation to try new things and new challenges.

Cultural fit

Every organization is different. I will conduct cultural fit interview based on situation. If company is fast paced then I will pick a candidate who likes such an environment.
If company is stable and not looking for much innovation then will pick a candidate accordingly.

Love for company's product offering

It best to see if they are interested in company's product offering. If someone hates the company's product then there is no point hiring them. Their work will reflect it in the product. If they are not doing it with love then why affect quality of the product.

----------------------------------------------------------------------------------------------------------------------------

How do you know an industry is ripe for disruption?

These points can provide some insights into it. But they are just the pointers, you will have to do some research to strengthen your insights.

Power is consolidated
Consumers are using outdated Technology
Business pratices aren't changing despite -ve feedback
The research backs you up
Reference: Link