Interview experience: AI Engineer

Interview experience: AI Engineer

Recently I gave an interview for the role of an AI engineer at a firm, it went downhill after a bit, but let's see the how it went.


I applied for the role on LinkedIn about 2 months ago, recently I got a call from the HR asking me if I was ready for an interview. I was pumped because the HR reopened the role after 2 months so I was expecting to be a favoured candidate.


Round 1

I join a google meet, where I'm interviewed by a senior IT professional, he gave me a heads up about the roles, responsibilities and the type of work they do. I'm was amazed with their deep research work, leaving me more excited.


The interviewer : "Tell me about your personal background, work experience, familiarity with Django, vector databases, LLMs, tensorflow, etc."

Me : I've responded honestly that I know some of them but not everything.


The interviewer : (responded positively and said) "Okay we can move to the next round, but before that i want you to look into Knowledge graphs and Cypher. These topics will constitute task 1 and the other task will be revealed at the spot."

Me : "Sure sir, I'll look into them. Thank you."

The interviewer : "Okay, then let's meet again. HR will send you the invite for the next round"


Drama starts, preparation for round 2

As the interview highlighted some topics, I've started learning & working with them. As a part of it I've build an application with Neo4j & Gemini using Langchain to interact with the graph database directly via an interface (LLM).

Round 2, Task 1

The interviewer : "Can you share me your SSH key?"

Me: "I don't think I can do that.."

The interviewer : "Haven't you worked with remote servers before?"

Me: "I did, but i didn't need to configure anything as the cloud providers provisioned & set them in place."

The interviewer : "Then we can't proceed with the first task, let's move on!"

Well this was my strike 1, not knowing much about remote servers.

Round 2, Task 2

The interviewer : "In this task we'll test your python skills, firstly create an virtual environment"

Me: "Okay."

I ran the command "python3 -m venv virt" in the terminal, creating a virtual environment named 'virt' in the root folder.

The interviewer : "Activate it"

Me: "source /bin/activate" in windows ??.

The thing is that I was working lately so much on ubuntu that I forgot the commands in windows. That was an embarrassing moment, in the end I couldn't activate it. Making it strike 2, my confidence went real low after this.

The interviewer : "Okay we can move on, generate me 100 records of a random name, timestamp, date in 2024 & an integer."

Me: I wasn't too familiar with the 'datetime' library as I rarely ever used it. My response was

import random
names = ['A','B','C','D']
random_name = random.choices(names)
random_number = random.randint(1, 100)        

I couldn't move further so I requested to check the documentation, to which the interviewer agreed. But the documentation was useless when you don't know what to search for. At the end strike 3, time up!


My conclusion

  • Interviews are stressful, we need to be composed under pressure or it will make us even less productive (like it did in my case)
  • Be honest and confident in yourself, If we are just confident in ourselves the pressure will not reach us, making it less stressful. If we try to attempt something that is something not familiar of, that just piles up the pressure. (Tip: Don't do it, it's a horrible situation to be in)
  • Focus more on the basics, never did I think that I'll fail a question on 'datetime', so don't be reckless and practice all the basics regularly.

Unanswered questions

  1. "Enter these records into a vector database"
  2. "Query the vector database via an LLM"


I'll publish a solution guide to this interview in the upcoming editions, stay tuned!

P.S. Thanks for reading, hoping this article helps. Love to hear your thoughts on this

AI Unplugged: Purely Human Perspectives        


Balusu Bhanu Prakash

AI & Data Science Professional | IIT Hyderabad M.Tech '26 | IIIT Dharwad B.Tech '24

11 个月

Comment your such experiences down below! So we can heal together ????

要查看或添加评论,请登录

Balusu Bhanu Prakash的更多文章

  • [Prototype] Creating GraphRAG before it was a thing

    [Prototype] Creating GraphRAG before it was a thing

    LangChain recently launched a new feature related to knowledge graphs named GraphRAG. This enables your agent to…

  • Learner's Choice: Qdrant vs Milvus

    Learner's Choice: Qdrant vs Milvus

    From the previous editions we explored a bit about vector databases. In this article, I want to compare two top…

  • MLOps series: What are 'Github Actions'?

    MLOps series: What are 'Github Actions'?

    MLOps is an extension of DevOps, mainly focusing on building, deploying, & maintaining machine learning (ML) models…

  • What is 'pymilvus'?

    What is 'pymilvus'?

    As an extension to the first edition What is Milvus?. Let's see how we can create a new milvus database using…

  • What is Milvus?

    What is Milvus?

    Since artificial intelligence (AI) is expanding rapidly in modern times, we must stay up-to-date with these new…

社区洞察

其他会员也浏览了