These are questions that will 100% come up in the MINT interview
AI — how you use it, etc.; you need to give a solid answer. You need to show interest in Python, that you read architecture books. Questions about microservices (Saga pattern, how to test). Celery. Kafka topics, partitioning. DDD is a must (Repository pattern, Unit of Work). Dependency injector (the whole project uses it — what it is, what it’s for, what it does). How to handle retries (if a microservice dies). You need to know async (multithreading, multiprocessing).
I’m calling you.
7/30 3:34 PM Julia Starygina started a meeting.
DDD is a must (Repository pattern, Unit of Work).
Basically, everything you’ve written here is correct.
mint S.P.A — Advertising Resource Management (ARM)… (already in English)
It’s a large marketing platform: a business comes in with money, tops up its balance, and then connects various ad campaigns on the platform. The platform lets you track spend, results, etc.
The questions will lean toward APIs, microservices, AI agents, LangChain, how to test agents, knowledge graph.
Still, the emphasis is on the AI part.
They mostly asked only about AI (RAG, token optimization, caching, MCP, tool-call optimization — if we have hundreds of tools, how to optimize context usage).
Technology list
Languages & runtime Python, SQL, Linux, Go, Scala, TypeScript
Web / API FastAPI, Django, REST APIs, microservices, distributed systems
Async & concurrency AsyncIO, multithreading, multiprocessing
Messaging & background processing Celery, Kafka (topics, partitioning), event-driven architecture
AI / LLM core LLMs, foundation models, LLM APIs, RAG, embeddings, vector search, vector databases, knowledge graphs, context management, prompt versioning & testing, structured outputs, tool calling, agent workflows, multi-step orchestration
AI frameworks & protocols LangChain, LangGraph, LlamaIndex, MCP
LLM production concerns Evaluation harnesses, regression testing, CI quality gates, guardrails, model-provider abstraction, retries, fallbacks, caching, rate limiting, graceful degradation, token/cost optimization, latency & throughput tuning, tool-call optimization at scale (hundreds of tools)
Architecture patterns DDD, Repository pattern, Unit of Work, Saga pattern, Clean Architecture, Hexagonal Architecture, dependency injection (dependency-injector)
Testing Unit, integration, contract, end-to-end, API testing, agent testing
Cloud & infra AWS, Docker, Kubernetes, Terraform
Observability Logging, tracing, metrics, alerting, auditability, AI observability platforms
Data ETL, data pipelines
Domain AdTech / MarTech, Advertising Resource Management (ARM)
The signal across all three note-takers is consistent: the AI half carries the most weight, but DDD + dependency injection came up twice independently, so those are the non-AI items most worth drilling.