LLM Interview Questions: Evaluation and Release Gates
A 12-question study guide to measuring LLM quality, building trustworthy evaluations, and deciding when a prompt or model change is safe to ship.
Search
30 results for “evaluation”
A 12-question study guide to measuring LLM quality, building trustworthy evaluations, and deciding when a prompt or model change is safe to ship.
A practical design for testing agent outcomes and tool-use trajectories offline, then turning the results into trustworthy release gates.
A medium-scale AWS blueprint for building a production RAG pipeline, from document ingestion and hybrid retrieval to reranking, evaluation, and operational readiness.
Prepare for AWS interviews with 14 questions on multi-account access, IAM policy evaluation, organization guardrails, audit trails, and operational design.
Choose prompting, retrieval, fine-tuning, or a combination by separating knowledge from behavior—and measuring their full cost and quality trade-offs.
A practical AWS blueprint for a small retrieval-augmented generation service: one API, PostgreSQL with pgvector, managed models, and an operational plan sized for a small team.
Q14 — EF Core Performance — Compiled Queries, Split Queries, Raw SQL Level: Senior | Topic: Entity Framework Core Performance Bottlenecks in EF Core Every LINQ query goes through an expression tree compiler that translat...
Q8 — JWT Authentication + Policy-Based Authorization in ASP.NET Core Level: Senior | Topic: ASP.NET Core JWT Authentication Flow The client authenticates once (POST /auth/login) and receives a signed JWT access token. On...
A production-minded AWS blueprint for a single-agent customer support assistant that searches a PostgreSQL knowledge base through constrained tools.
A production-minded AWS blueprint for a LangGraph research assistant that delegates bounded research, checks evidence, and returns traceable answers.
Prepare for Azure interviews with 12 practical questions on management groups, landing zones, policy, identity, safe rollout, and operating governance at scale.
A small-team AWS blueprint for turning structured logs, external uptime checks, traces, and a focused dashboard into a practical production feedback loop.
Here's a comprehensive breakdown of all 10 features, from junior through to principal engineer level. I'll build a visual architecture overview first, then walk through each. Now let's go deep on each feature, level by l...
Prepare for SDLC and DevOps interviews with 13 practical questions about immutable artifacts, deployment strategies, release gates, rollback, and production feedback.
A medium-scale AWS blueprint for instrumenting Node.js and .NET services with OpenTelemetry, propagating context across a queue, and operating traces, metrics, and logs.
A medium-scale, AWS-based blueprint for reliable ingestion, warehouse modeling, orchestration, delivery, and operations with Airbyte, Snowflake, dbt, and Dagster.
Q3 — Explain the C# Type System: Records, Nullable Reference Types, and Pattern Matching Level: Mid-Level | Topic: C# Fundamentals Records (C# 9+) Records are compiler-synthesised types that provide value equality , immu...
This design assumes you want reliability, observability, security, compliance, scalability, and future flexibility. 1. Executive Architecture Summary A production AI-agent platform should not be just: User → LLM → Respon...
A production blueprint for a medium-scale SaaS on AWS, covering tenant isolation, transactional messaging, delivery pipelines, observability, cost and recovery.
A practical study guide to defining useful SLOs, designing burn-rate alerts, and explaining the telemetry choices behind reliable production systems.
Twelve practical interview questions on selecting, validating, operating, and evolving a technology stack without treating familiarity or fashion as evidence.
Prepare for interviews on software supply-chain security, from dependency pinning and SBOMs to signed provenance, deployment policy, and pipeline observability.
A small-team architecture blueprint for a clinic booking app, from PostgreSQL conflict safety and App Service deployment to observability, recovery, and cost.
Caching can cut repeated work and protect dependencies, but only when keys, freshness rules, and failure behavior are designed as carefully as the hit path.
A small-team blueprint for giving colleagues useful, governed analytics with Metabase and PostgreSQL—without introducing a warehouse, a data platform, or a second full-time job.
Vertical scaling gives one machine more capacity; horizontal scaling spreads work across machines. The right choice depends on bottlenecks, state, failure tolerance, and operational cost.
Design traffic steering as a versioned control plane that publishes safe routing policy to fast, independent data planes, with explicit health, locality, capacity, and failure behavior.
Q11 — Eager, Lazy, and Explicit Loading — Differences and the N+1 Problem Level: Mid-Level | Topic: Entity Framework Core The Three Loading Strategies Eager Loading ( Include() ): Related data is loaded in the same query...
Q22 — OWASP Top 10 — How Each Applies to .NET APIs with Mitigations Level: Senior | Topic: Security The Most Critical Risks A01 — Broken Access Control: Never trust client-sent IDs. Always load the resource then verify o...