How to be better at communication via talking with gpt
Hey guys,
I want to get better at communication in interviews so I started giving interview to chatgpt
( he know me well )
i have given many real interview & found while my engineering skills aren't bad but my communcation skills sucks.
here is my voice chat with gpt :
Can you tell me what i can do better
#############################################
Introduction
Interviewer: Tell me about yourself.
Me:
Hi everyone, I am Manas Anand. I have been backend engineer at Raffles.ai for past two years, mostly working with AI backend system. I have architected microservices that handles real-time AI inference, storing data in PostgreSQL database for user state, and pushing multiple changes to front-end. My strengths are in optimizing for scale, asynchronous pipeline, and collaborating across the stack. Apart from academics, on weekends I really enjoy playing guitar and sometimes football.
Feedback
Rating: 8/10
Improvements
- Say “AI inference” instead of “AI interference”
- Maintain consistent grammar:
- “microservices that handle”
- “backend engineer at Raffles.ai for the past two years”
- Strong structure and confidence overall
DRY Principle
Question
Explain DRY with a practical example.
My Answer
DRY stands for Don’t Repeat Yourself. In our organization, we were using OCR processing heavily across multiple pages. Instead of rewriting the OCR logic every time, we made a modular OCR function that could be reused in different places.
Now multiple services can call the same OCR pipeline, which improved readability, maintainability, and centralized the business logic.
Feedback
Rating: 7/10
Improvements
- End with measurable impact
- Example:
- “This reduced duplicate logic and simplified maintenance.”
YAGNI Principle
Question
Explain YAGNI with a practical example.
My Answer
YAGNI stands for “You Aren’t Gonna Need It.” It prevents over-engineering for scale that may never happen.
In our organization, we needed an AI agent for a specific workflow. Some engineers suggested creating multiple sub-agents, but the use case was not complex enough.
I benchmarked a single well-prompted agent with structured outputs, and it performed well enough. This reduced complexity, simplified logging, and kept the codebase easier to maintain.
Feedback
Rating: 6/10
Improvements
- Keep answers tighter
- Clearly state:
- Problem
- Decision
- Result
SOLID Principles
Single Responsibility Principle (SRP)
My Answer
SRP means a class or function should have only one responsibility.
In our OCR pipeline, the first version handled OCR, JSON formatting, and logging all in one class. I refactored it so the OCR service only extracted text, while logging and JSON parsing were moved to separate modules.
This improved readability and modularity.
Feedback
Rating: 7/10
Improvements
- Shorten explanations
- Keep one clear responsibility statement
Open/Closed Principle (OCP)
My Answer
Open/Closed Principle means software should be open for extension but closed for modification.
For example, in a logger system, we may initially support only info and error logs. Later, if we want to add debug logs, we should extend functionality without modifying stable existing logic.
This reduces the chance of breaking existing systems.
Feedback
Rating: 7/10
Improvements
- Say:
- “Add new behavior without editing stable existing code.”
Liskov Substitution Principle (LSP)
Correct Explanation
A subclass should be replaceable for its parent class without breaking expected behavior.
Example
A classic violation is:
Rectangleclass allows independent width and heightSquaresubclass forces both values to remain equal
This breaks client expectations and causes bugs.
Interface Segregation Principle (ISP)
My Answer
ISP means clients should not depend on methods they do not use.
For example, instead of one huge Vehicle interface, we can split interfaces like:
EnginePoweredSwimmable
This prevents unrelated classes from implementing unnecessary methods.
Feedback
Rating: 6/10
Improvements
- Define the principle first
- Then explain interface splitting
Creational Design Patterns
Factory Method Pattern
My Answer
Factory Method is used when object creation depends on runtime conditions.
For example, if we have multiple car types like Audi, BMW, and Ferrari, instead of hardcoding object creation everywhere, we use a factory that returns the correct object based on configuration or input.
This improves maintainability and separates creation logic from business logic.
Feedback
Rating: 7/10
Improvements
- Emphasize:
- “Encapsulating object creation”
Singleton Pattern
My Answer
Singleton ensures only one instance of a class exists.
A common example is database connection management. Without Singleton, multiple unnecessary connections may exhaust the connection pool.
Singleton helps centralize resource management.
Risks
- Hidden global state
- Harder testing
- Concurrency bottlenecks
Feedback
Rating: 7/10
Builder Pattern
My Answer
Builder Pattern is used for constructing complex objects step-by-step.
Telescoping constructors become difficult to read when there are many parameters. Builder pattern makes object creation more readable and maintainable.
Example:
User user = new UserBuilder()
.setName("Manas")
.setTimeout(30)
.enableCache(true)
.build();
Feedback
Rating: 8/10
Prototype Pattern
My Answer
Prototype Pattern is used when cloning existing objects is more efficient than creating new ones.
In game development, character templates can be cloned and then slightly modified instead of recreating everything from scratch.
One important concern is shallow copy vs deep copy.
Feedback
Rating: 8/10
Overall Interview Feedback
Strengths
- Strong practical understanding
- Real-world engineering examples
- Good instinct for scalability and maintainability
- Comfortable discussing backend architecture
Main Areas to Improve
1. Sentence Precision
Your biggest issue is not knowledge — it’s sentence construction.
You often:
- restart sentences
- repeat ideas
- lose grammatical structure midway
Fix
Use shorter sentences.
Instead of:
>
Say:
>
2. STAR Structure
You are partially following STAR naturally.
You consistently:
- explain context
- explain action
But you sometimes skip:
- measurable result
Better Structure
- Situation
- Task
- Action
- Result
Always end with:
- reduced latency
- improved maintainability
- simplified debugging
- reduced duplicate code
- improved scalability
3. Pace & Delivery
Your pace changes depending on confidence.
Pattern Observed
- Strong topics → calm pace
- Searching for words → very fast pace
Fix
Pause intentionally.
Use transition phrases:
- “In short…”
- “The main idea is…”
- “The outcome was…”
This buys thinking time and sounds confident.
English Improvement Strategy
Best Practice (30 mins/day)
1. Record Yourself
Take one interview question daily.
Record a 1-minute answer.
2. Rewrite It
Rewrite into:
- 3 clean sentences
- simpler grammar
- fewer filler words
3. Speak Again
Repeat until it sounds natural.
Final Overall Evaluation
| Area | Rating |
|---|---|
| Technical Understanding | 8.5/10 |
| Practical Examples | 8/10 |
| Communication Clarity | 6.5/10 |
| Confidence | 7/10 |
| Structure | 7/10 |
| Interview Readiness | 7.5/10 |
Final Advice
You already think like an engineer.
Now your next jump is sounding structured and concise under pressure.
If you improve:
- pacing
- sentence clarity
- concise STAR delivery
…you can perform very strongly in senior backend interviews.
Ps: I want to share the thread but it's not available for voice
