The discussion started with a brief introduction about my background and recent projects. The interviewer then quickly moved into a **deep dive on my Deloitte project**, asking detailed questions around:
* RAG pipeline design
* Handling large PDF documents (10–80 MB, multi-page parsing)
* LLM chaining and orchestration
* Failure handling (retries, fallback, validation layers)
* State management in multi-step workflows
They were particularly interested in **production challenges**, like:
* Handling LLM hallucinations
* Ensuring deterministic outputs
* Latency and scaling (1000 concurrent requests scenario)
* Async vs queue-based architectures
---
* **Design thinking**
* Clean structure
* Handling edge cases
---
### 👍 **What I Liked**
* Interview was **very practical and scenario-based**
* Focused on **real production challenges**, not textbook questions
* Interviewer was knowledgeable and pushed for **depth of understanding**
* Good discussion around **AI system architecture**
* Writing clean Python code for workflows
* Explaining system design clearly
* **Difficulty Level:** Medium to High
* **Focus Area:** Applied GenAI + Backend Engineering
* **Recommendation:** Great opportunity for engineers with hands-on experience in **LLMs, RAG, and production AI systems**
Overall, a **great learning experience** and a solid interview process for anyone serious about working on real-world AI systems.
