Current Status:
Position: Lead ML Engineer at Informatica
Location: Bangalore
Total YOE: 7.5

Recruiter reached out to me after I applied via LinkedIn

Hacker Rank Online Assesment Test:
First was a 90 minute hackerrank online assesment test which was based on data manipulation. The problem statement was related to reading a large chunk of file and manipulating it in a certain way. It didn't involve much use of data structures and algorithms. Some experience with python scripting and data structures like hash map should be good to solve it. Had to pass 8 out of 10 testcases. Difficulty level was easy to medium. Any engineer with some experience with python scripting should be able to do it.
I cleared the assessment and got a call from recruiter for interview rounds the next day.

First Interview Round: System Design (ML)
Was asked to design a system for Image Classification starting from data collection, preprocessing, model finetuning and deployment. Had discussion on various components as to how data will be structured/stored and the entire pipeline/design will be laid out including the various APIs that will be required. There were follow up questions on the deployment and the scaling strategy.
This round didn't went too good for me. I was not expecting a call forward. But they did after few days.

Second Interview Round: DSA and ML
The round was divided into two sections. ML and DSA. First 30 minutes were ML. Interviewer asked questions from CV projects and finetuning models and traditional machine learning interview questiosn like bias-variance tradeoff, underfitting/overfitting, transformers architecture, peft finetuning etc. Since my projects involved transformers and PEFT that's why questions were asked on those topics.

Then two DSA problems were asked:
First problem was on bit manipulation, given an integer need to convert it to ipv4 address. Difficulty level was easy to medium. I was able to solve it.

Second problem was on linked list, given a linked list, remove the nth element from last. It's a standard problem. I was able to solve this with ease.
This round went very well for me overall.

Third Interview Round: DSA and ML (with Hiring Manager)
This round was also divided into two sections. ML and DSA. First 30 minutes were ML. Manager gave a sense of what team does and sort of use-cases they are working on. Then delved into ML questions. Again, was very much the same as previous round. Questions were asked from my projects. Transformers Architecture, Discussion on contrastive learning vs binary cross entropy loss, Finetuning a decoder model (PEFT vs full finetuning vs pretraining). There were few questions on traditional ML as well. (very standard ones like precision/recall metrics, ROGUE/BLEU etc).

Then the interviewer threw a problem statement of reading a very big dataset file. I was expected to come up with approach and write the pseudo code for it. Basically we had discussion on using generators in python to read it chunk wise and then what will be the chunking strategy to ensure duplicates fall under the same chunk. I was able to come up with the approach interviewer was expecting and was able to write the code as well.
Overall, this round also went really good.

Overall, all the interviewers were very professional and friendly at same time. Made me feel comfortable during the entire interview process.

Right after this round, HR contacted me saying you've cleared all rounds. We will roll out the offer. However the offer was actually rolled out after about 15-20 days. I lost hope at one point that they will not roll it out since they may have got better candidate but they fortunately they did.
I will write the compensation post soon.