tl;dr: No offer but good experience, except for one interviewer.

1st phone screen was with a recruiter who asked questions on my overall background and interests. 2nd phone screen was with a senior manager who focused on my recent experience, interests, and skills.

Onsite:

I had 4 technical rounds and 1 non-technical round that focused purely on behavioral and team management aspects.

The 4 technical rounds were as follows:

• Questions on design and architecture of distributed systems.
• Mix of system design, previous projects, and management style discussions.
• Focused on a previous distributed systems project, with a large data set problem and a database-related question.
• Whiteboard coding round with a slight mix of management-related questions.

I had a rather unpleasant experience with one individual. This person had a condescending bearing throughout the interview. The nadir came when he asked me a database-related question which I answered as best I could, unfortunately to be met with outright derisive laughter on his part. I took that as my solution being incorrect in his eye and presented a different approach only to elicit the same odious reaction from him. I let it pass at that moment, but later on verified that my original answer was actually exactly how engineering teams at many tech companies solved the problem. Of course, there are alternate ways to tackle this problem as well, and perhaps the interviewer had one of those ways in mind. Nevertheless, his behavior was quite uncouth and unprofessional.

One of the cultural values at Credit Karma is supposedly "EMPATHY - We seek first to understand, then be understood." If only they would first ensure their employees understood their own values before spouting them in emails to candidates.

To be fair, I should point out that this particular individual's first language and medium of instruction in his native country is not English, so perhaps he had some difficulty in understanding Credit Karma's values. I did have some trouble communicating with him.

As I was walking him through one of my previous distributed systems projects (scale of billions; RPS in millions; globally distributed), he made supercilious remarks quite a few times insinuating how trivial it was and how easy it would have been for him to work on it, while ironically failing to grasp a lot of the fundamental concepts. I thought perhaps he had difficulties in comprehension due to linguistic issues and often had to repeat things in a slow and simplified manner just so we could try and find some common ground.

Other than this one individual, I enjoyed speaking to the rest of the panel. They were all technically sound within the context of their current work at Credit Karma and perhaps their immediate prior workplace. Most questions were focused around problems they had encountered and/or resolved.

The interviewers were comfortable with the distributed systems and microservices domain from an application level, but they seemed foundationally unaware about some of the core concepts. For instance, a couple of them (including the aforementioned obnoxious person) who focused on distributed systems questions had no idea about Paxos or Raft and had not heard about distributed consensus or leader election algorithms.

Based on the body language of some of the interviewers, it seemed that if the given answer did not align with what the interviewer had in mind, that would very likely go against the candidate, irrespective of the trade-offs between the different approaches.

One of them did not seem very happy when I presented an approach for a service discovery question that used an open-source Apache technology that he had clearly not heard about. Or, for that matter, when I answered a question about dealing with very high computational load on a single instance of a microservice by presenting a solution using horizontal scaling and asynchronous distributed computation patterns. Or, when I pointed out that simple round-robin load balancing could very well lead to undesirable loads on certain instances.

This broadly agreed with my observation about the interviewers' technical strengths being drawn more from specific usage experience and prior application patterns rather than the conceptual aspects of distributed systems.

I failed to answer a relatively easy question on security, but the solution just slipped my mind at that time, and faltered on another broadly related to MapReduce, although I did come up with the right algorithm. And of course, my answer to the database question clearly was very amusing to that one particular interviewer.

All said and done, overall I gained a few things from the interview experience, especially since I was interviewing after more than a decade, so thanks to Credit Karma for the opportunity. They have some interesting problems at hand as they scale their business, user base, and organization, and I wish their efforts well.

Questions on security, scaling, microservices, MapReduce, database, and management style.