here is the detailed post, how i got selected as an ml engineer:
so this whole thing started with a viral post i wrote before, which caught attention of a senior professional he reached me out to me on twitter(and also found me on LinkedIn). He invited me to interview for his startup, and that’s how it all began.
so this whole interview consists of 3 Rounds
Round 1 (1 hr - 1: 30 hr):
ml models - their workings, principles and maths intuition behind model
core maths topics - vectors, linear algebra, correlation, bayes theorm, probability and properties
situational question - 'how you will teach ml to first year clg students ?'
general programming questions - python, oops, java
then we moved to dl - perceptron, activation functions and also the selection of activation function on different use cases, working of CNNs - ocr, pytesseract, icr
mp4 data decryption, RNNs, LSTMs
then we moved to nlp - text preprocessing, bag of words, word to vectors, language models (bigram, trigram) then we deep dive into transformers - full detailed explanation of transformers- encoder, decoder working, QKV vectors, normalization, masking
difference between LM and transformers, BLEU score, also he was doing cross questions which i was explaining them in positive way
then we discussed about my projects - discussed every project what, why, how ? so many cross questions like why you did this, instead of we can have done this ?
then previous working experience, iit reasearch, and abt my college studies
then we moved to vector databases - detailed questions like cosine similarity and different vector db platforms then we went to agentic ai - agents, frameworks and langchain, and rag pipelines, here i found some difficulty but it was okay interviewer was helping me too.
this was Round 1
and he told me to note some things for round 2 - mcp protocols, cloud technologies, agents, and adk platform
so after 2 - 3 days, i reverted back to him and said that i am ready for Round 2
Round 2 (30 - 40mins):
here we started with again vector databases, agents and llms
vector databases - dimensions, properties, how to connect them
agents - detailed questions like wht, how , why ?
google agent developement kit - types of agents, databases, model context protocols, llm agents, function tools - built in tools , agent tools, workflow of agents, db how to connect to agents, multi agents, sequential agents, loop agents, parallel agents, session - state - and runners, types of sessions - memory, db, vertex ai
then he gave me an assignment to make a bot combine all things like ml, nlp. agents
this was Round 2
then after 4 to 5 days i again reverted back to him and i said i am ready for Round 3
Round 3 - demo of that bot
full explanation of that project - wht technologies used, why used, and use cases of each agents
showed them the full working of the project and so many cross questions they asked, i explained them each of them clearly and softly also i was running that whole project on their prompt whatever they were saying - so i already made a robust bot and it performed well
and this was the last round so ceo was also there, he asked me some questions, about my learning, my past experiences, and my current studies
yes this was the Round 3
after 2 days i got the selection letter
SELECTED as an ml engineer !!