Python

What work format are you looking for: office, hybrid, or remote?

How did you interact with ML/AI engineers and what model or AI solution did you implement?

Are you familiar with request processing products, anti-fraud, AML, credit conveyor, and sanctions screening?

5. Payment reliability: "How was the issue of idempotency solved in gRPC?" Interviewer’s goal: Payments — the most dangerous area for duplicates (Double Spending). Possible question: "The payment request went through gRPC but failed due to timeout (DEADLINE_EXCEEDED). Did the money get deducted — unknown. What does the payment module do?" How to answer: "In a financial transactional context, any repeated request must have an Idempotency Key: 1. In the metadata of the gRPC request (metadata=((‘x-idempotency-key’, key_uuid),)) we always transmitted a unique transaction key. 2. When receiving DEADLINE_EXCEEDED, the client could safely Retry with the same key. 3. The acquiring service checked the presence of this key in its database / Redis and either returned the status of the already processed transaction or processed it for the first time."

What was your team like, who did you interact with, how did you work, and what stack did you use?

Why did you leave your previous job?

What are the requirements for storing history, message order, search, encryption, availability, message region, and deletion?

Which local model was used: Qwen, DeepSeek, or another?

Response to Python developer vacancy

Design a messenger.

What are you looking for now and what is key for you when choosing a company?

Did you use AI agents or tools for code generation, such as Codex?

Tell us about the most complex technical or product challenge you faced: what was the problem and how did you solve it?

Tell us about a failure at work: what happened, why, and what did you learn?

Design a scalable messenger: only personal chats, sending text and images, online and offline delivery via push notifications, statuses "sent/delivered/read", chat history, and online/offline presence status. Constraints: up to 10 million users, average load up to 50k messages/sec, peak — up to 100k messages/sec, messages must not be lost, system must scale horizontally; P99 online delivery — up to 400 ms.

Besides Python, which other languages have you used or are you interested in now?

What AI tools besides the local model did you use?

What functional and non-functional requirements should be set for a messenger: personal chats, message statuses, files up to 5 MB, push notifications, up to 10 million users, 40–50 thousand RPS on average and up to 100 thousand RPS at peak?

Do you agree to abandon end-to-end encryption to support server-side full-text search on messages?

Several questions have arisen Is it necessary to write about school education? Should I indicate anything about computer skills? If yes, should I just put it in the first column? Because the table is clearly a bit off.