Sobes.tech

How to add a specific business metric to an application — what needs to be done, what it would look like in code?

132

Tell us about the services and work at your last job — pick a couple of features and tell how you solved them from a technological perspective, what patterns you used, what difficulties and alternative approaches there were.

128

Did the ML service perform data enrichment from external sources and run the model internally, or were these functions divided between services?

128

How have you worked with Kafka? What does Kafka partitioning provide in terms of performance?

127

Tell us about your experience integrating ML/AI APIs — have you worked with OpenAI or other models?

127

In Kafka, there is a topic with 10 partitions. We added another consumer replica, but the performance did not increase. Why is that and what should we do?

122

Describe the technical architecture of the scoring platform component interacting with the ML model: service separation and responsibilities, communication between them, and your personal contribution.

120

Tell us about the interaction of services via REST and Kafka — what were the use cases for each approach?

119

What happens to a Kafka message after being read by a consumer: will it continue to be stored?

118

When starting a Flask application with `python main.py`, a warning appears in the console that it is not for production. What does this mean, why does it appear, and how to deal with it?

118

How did the two services interact via Kafka: what happened at the input and output, how did the request pass through the pipeline to the decision of approval or rejection?

118

How was the caching layer organized, and what optimization features were used for your specific case?

117

How many models were used in the scoring process? Were there non-model, empirical scoring rules?

117

What would you like to find for yourself now, what is interesting and important when choosing a future employer?

115
/4