Page 30 of 129
Cursor engineer Lauren Tan (poteto) says she merges roughly 2,000–2,500 PRs a month, a self-reported figure no one has audited. Her method is three moves: build a verification skill first, turn repeated mistakes into lint and type constraints, and sample-review after merge. The method transfers if your domain can be verified by code.
Ask about sun with Does this wall get the sun…; drying time with How long does it take to dry…; rules with Do I need to register…; rentals with I'd like to rent…; length with long enough; partners with I'm looking for a climbing partner; routes with Can you recommend…. More rope is Slack!, less rope is Up rope!, and in a gym, asking the belayer to hold you and lower you from the top is Take!.
Positioning: I work on improving…, My work covers…. Changing requirements: changed five times in four months. Persuading with data: that convinced the team to…. A measured result: took 29 minutes. A mistake: I changed … but didn't update …, We only found out … after …. An honest gap: To be honest, we don't have … yet.
Expertise: I have experience in / as. Current job: I currently work for a company that…. Ownership: I was responsible for + -ing. Results: reduced … from A to B. Difficulty: The hardest part was + -ing. Reflection: I underestimated how…. Change: Since then, I've always…. For your own questions, ask how the team divides up the work and what the goals are for the first three months.
My friend understood me, but the same six things kept going wrong: use Have you + past participle for 'so far' questions; apply to a company, apply for a job; interview + a person means you are the interviewer; other takes a plural noun while others stands alone; because takes a clause and because of takes a noun; adjectives need be, and sentences need a subject.
Report finished work with the past simple plus a time; say I'm still waiting for when you are waiting; use by for a deadline; say I'm stuck on for a blocker; ask with Could you take a look at… when you have a moment; soften disagreement with I'm not sure about… and then suggest with Could we…; estimate with This will probably take….
Uber runs one MCP Gateway for 800+ MCP servers and 5,000+ tools: a control plane crawls internal APIs into disabled-by-default tools, a data plane handles protocol translation, authorization and redaction, and Omni MCP, Response Projection and Code Mode keep context small at scale.
Pi 1.0 and the experimental Pi Durable package shipped together in early October 2026. Pi Durable turns every model request, tool call and compaction into a task that writes a checkpoint, so a killed process resumes from its last checkpoint; Cloudflare followed with PiHarness on Durable Objects. The key design decision is classifying tools as safe or unsafe to replay.
Agent interviews keep returning to eight topics: the agent loop, structured output, function calling, large tool sets, memory, design trade-offs, MCP versus A2A, and prompt caching. This post strings them into one line (concept, mechanism, how to answer) and corrects when the 2026-07 MCP spec changes actually landed.
Four system design questions that come up often (enterprise RAG, an LLM serving stack, a customer-support agent, online A/B testing with a data flywheel), three Python problems (dynamic batching, sampling, backoff retries plus a token bucket), and a STAR skeleton for behavioral rounds. Each design question has a speakable answer and a self-check list, and all code was actually run.
Seven LLM engineering topics that come up often in interviews: LoRA/QLoRA and fine-tuning vs RAG, RLHF/DPO/GRPO alignment, inference optimization (PagedAttention gives vLLM 2-4x throughput), sampling and decoding, LLM evaluation, the OWASP LLM Top 10 (2026 edition), and hallucination. Each section covers the concept, the mechanism, and how to answer.
Interview follow-ups usually land on mechanisms: why divide by √d_k, how AdamW differs from Adam plus L2, why a high AUC still isn't shippable. This post walks through concept, mechanism and how to answer, then ends with 3 runnable numpy exercises.
Prompt engineering shapes a single instruction, context engineering decides what the model sees right now, and harness engineering wraps the whole non-deterministic system. Bsharat's 26 principles, Breunig's four context failures, and OpenAI's roughly one-million-line, zero-handwritten-code harness experiment each map to one layer. Evaluation is the piece teams skip most, and it belongs in a CI gate that blocks silent regressions.
Answer RAG interview questions in four steps: the three-stage pipeline, where each stage fails, which of nine variants patches which hole, and the 6 kinds of problems RAG cannot solve. Self-RAG, CRAG and Adaptive-RAG add judgment; Graph RAG and KG-RAG change the index; CAG skips retrieval; Agentic RAG and Reasoning RAG hand decisions to the model.
AI Engineering from Scratch is a free, MIT-licensed course on GitHub with about 63k stars: 20 phases and 523 lessons, running from linear algebra, classical ML, and deep learning to LLMs, agents, and AI safety. Each lesson builds the algorithm by hand, then repeats it with a framework. The early math and deep-learning-core phases are worth following step by step; skip the Phase 10 SFT/RLHF/DPO lessons, and treat the rest as a map of topics.
Japan has no official supermarket ranking, but stores fall into three tiers by positioning: Kinokuniya and Seijo Ishii for gifts and imported food; Life, AEON, and Seiyu for everyday groceries; OK, Gyomu Super, and Lopia for bulk and low prices. At checkout, food is taxed at 8% and alcohol at 10%.
Read each company's own interview guidance first: Meta now expects candidates to use a built-in AI assistant, while Anthropic defaults to no AI. For question banks, pick ones that cite sources and are still maintained this year, such as the alexeygrigorev field guide, ombharatiya, and alirezadir. Only 2 of the 12 popular repos link sources for their questions, and some keep answers behind a paywall.
Audio representation, temporal relationships in video, and action-conditioned world transitions are different learning tasks. The earlier combined guide checked only audio at the end. This URL now serves as a reading map to one full guide each for Lectures 24–26; the Lecture 26 guide also covers the separate alignment deck.
Text models convert text into tokens; image models also need a computational representation of pixels.
Code Generation and Agents: A Reliable Test–Repair Loop | CMU 10-423 Lecture 23