Hello Reader,

Six months ago, an interviewer asking about Gen AI was a bonus round. Today it is standard. At AWS, Microsoft, Meta, JP Morgan, Verizon, and most large enterprise technology teams, Gen AI questions are showing up in every SA, FDE, and AI engineer interview regardless of the role's primary focus.

The candidates who answer these well are not the ones who have read the most blog posts. They are the ones who can connect the concepts to real architecture decisions and explain the trade-offs without getting lost in jargon.

Here are the questions coming up most often right now and the answers that actually land.

Tell me about an agent orchestration pattern you have used

The average answer

I have used Human In The Loop (HITL)

This answer is actually not average, but wrong!

Why wrong?

Pay attention to the word "orchestration." It specifically means coordination between multiple agents, not a single-agent feature. Human in the loop is not an orchestration pattern. It is a single-agent pattern, since frameworks like Strands or LangGraph already support pausing for human input without needing multiple agents at all.

Good answer

There are multiple orchestration patterns:

• Loop
• Sequential
• Hierarchical

Then pick one where you can go deeper on. Loop is a popular orchestration pattern

Agents repeat and improve their work based on feedback from another agent until an exit condition is met. This works best when tasks do not overlap or depend on each other.

The clearest example is generator and evaluator: one agent writes code, another agent reviews it against best practices and sends it back, and the cycle repeats until the review passes.

The real risk with this pattern is runaway cost, so you need to define limits in advance, maximum iterations or a token ceiling, so it does not spiral.

Some other exit conditions for loop pattern are max iteration count, a token budget, or passing certain quality checks.

Note: If you are going for FDE interviews, you also need to know the code implementation for the pattern. You don't need to tell the syntax, but you have to mention that, with Strands I will use Agent as Tool or Swarm to implement this. With LangGraph, I will use Graphs to implement this etc.

Can I insert everything in the context, and not use memory and save money?

This is a trap question. You can NOT replace memory with context.

The wrong answer here is telling the interviewer you just put everything in the system prompt to save money on memory infrastructure. That misunderstands what memory actually does.

Context that goes into a session isn't copied wholesale into long-term memory. It goes through an extraction process that pulls out specific things: user preferences, semantic facts, a summary of the interaction, and episodic memory, the complete reasoning trace, steps taken, and outcomes.

These are the reasons you can't replace memory with context:

• Memory is NOT just saved context E.g. User preferences, and summaries are derived from extensive conversations. If you don't use memory, LLM has to reprocess the context over and over again to derive the preferences, summaries etc. And processing tokens is expensive!
• Speed - Look up instead of processing same info over and over again. Memory is Keys + Values which means faster lookup
• Context full, you can resume the task in the next session (e.g chat sessions)
• Memory can be shared between Agents. For example, billing agent can pass the control to enrollment agent, and they both can share the memory to get the full lifecycle of the customer
• Memory is basically a Database. And databases have been around for decades. They are scalable, have granular controls, performance optimizable, and can have expiry control. You don't have these in context

Keep learning and keep rocking 🚀,

Raj

P.S - If you want to get an AWS Solutions Architect job without coding or learning every AWS service, the 10th cohort for AWS SA Bootcamp is launching on Oct 17th, 12 PM ET (Eastern Time) via live workshop. This program now includes our updated Gen AI - including FDE roles! Please register below:

Here’s what you get when you show up LIVE:

1. The myths keeping most people stuck - and what actually gets you hired as an SA - I've conducted over 300 SA interviews, so I know what I'm talking about!
2. How GenAI is reshaping the SA and Gen AI roles including FDE, and the exact AI concepts (RAG, agents, MCP, eval etc.) you need to speak fluently in interviews.
3. A first look at my new product feature, built to help you practice real-world, interview-relevant hands-on work instead of copy-paste tutorials.
4. Full bootcamp breakdown for Cohort 10, plus a special offer only for live attendees.
5. My exclusive Solutions Architect framework to prep you for today's job market! But if you’re not live, you won’t get it. No second chances.

And good news - it already worked for last cohort's students who secured cloud jobs in top companies, including at AWS, Microsoft, Google, JPMorgan, Reddit, and some of them didn't even have cloud experience 💰.

Spots are limited, so don't miss it!