# AI Startup & AI Lab Interview Guides

Source: https://www.techinterview.org/ai-startup-interview-guides/
Updated: 2026-07-15 · techinterview.org

related dataNew: our [AI-Startup Interview Difficulty Index](/ai-startup-interview-difficulty-index/) scores every company below from 1 to 5 on coding, system design, ML/research depth, take-home load, and speed — ranked hardest to easiest.
AI-native companies

AI-Native Company Interview GuidesInterview guides for the AI-native companies — the frontier labs and the startups building on them: OpenAI, Anthropic, Mistral, Cursor, Perplexity, Modal, Together AI and more. What each one actually tests: ML and systems depth, inference at scale, and shipping fast in hypergrowth.

**74** company guidesTokens, attention weights, outputthecachewasstalesoftmax(QKᵀ/√d)·V[A**Abridge**Abridge is a Pittsburgh-based clinical AI company whose product listens to clinician-patient conversations and turns…](https://www.techinterview.org/companies/abridge-interview-guide/)[A**Ambience Healthcare**Ambience Healthcare is a San Francisco company selling ambient AI documentation, point-of-care coding, and clinical…](https://www.techinterview.org/companies/ambience-healthcare-interview-guide/)[A**Anthropic**related dataOn our AI-Startup Interview Difficulty Index, Anthropic ranks 13th of 32 (3.4 out of…](https://www.techinterview.org/companies/anthropic/)[A**Assort Health**Assort Health builds specialty-specific AI voice agents that answer the phones at orthopedic, dermatology, urology…](https://www.techinterview.org/companies/assort-health-interview-guide/)[B**Baseten**Baseten is a San Francisco AI inference company that helps teams deploy and serve ML…](https://www.techinterview.org/companies/baseten-interview-guide/)[B**Black Forest Labs**Black Forest Labs interviews screen for deep expertise in diffusion and flow-matching models, generative image…](https://www.techinterview.org/companies/black-forest-labs-interview-guide/)[C**Cartesia**Cartesia builds real-time voice AI on state space models, so its interviews reward people who…](https://www.techinterview.org/companies/cartesia-interview-guide/)[C**Character.AI**Character.AI hires through a multi-stage loop that begins with a recruiter screen, moves to a…](https://www.techinterview.org/companies/character-ai-interview-guide/)[C**Cohere**Cohere interviews center on enterprise AI: expect to show you can build and ship real…](https://www.techinterview.org/companies/cohere/)[C**Commure**Commure is a General Catalyst-built healthcare operations company selling ambient clinical documentation, AI billing and…](https://www.techinterview.org/companies/commure-interview-guide/)[C**Contextual AI**Contextual AI builds an enterprise platform for grounded, retrieval-augmented generation it markets as "RAG 2.0,"…](https://www.techinterview.org/companies/contextual-ai-interview-guide/)[C**CoreWeave**A CoreWeave interview tests the skills behind running GPU infrastructure at scale, so expect rounds…](https://www.techinterview.org/companies/coreweave-interview-guide/)[C**Crusoe**Crusoe hires infrastructure-heavy engineers to build a GPU cloud that runs on power the company…](https://www.techinterview.org/companies/crusoe-interview-guide/)[C**Cursor**related dataOn our AI-Startup Interview Difficulty Index, Cursor ranks 3rd of 32 (4.0 out of…](https://www.techinterview.org/companies/cursor/)[D**Dagster**Dagster Labs ran a compact, practical loop: a recruiter chat, a technical screen with a…](https://www.techinterview.org/companies/dagster-interview-guide/)[D**Decagon**Decagon interviews focus on whether you can build and ship AI agents for customer support,…](https://www.techinterview.org/companies/decagon-interview-guide/)[D**Deepgram**Deepgram's interview process tests whether you can build and reason about production speech AI systems,…](https://www.techinterview.org/companies/deepgram-interview-guide/)[E**ElevenLabs**ElevenLabs interviews test whether you can build and ship AI voice technology at production scale,…](https://www.techinterview.org/companies/elevenlabs-interview-guide/)[E**Exa**Exa hires engineers who can build real search and retrieval systems, not recite textbook algorithms.…](https://www.techinterview.org/companies/exa-interview-guide/)[F**Factory AI**Factory is a San Francisco company building autonomous coding agents called Droids for enterprise software…](https://www.techinterview.org/companies/factory-ai-interview-guide/)[F**fal.ai**fal (fal.ai) is a generative-media inference platform, so its engineering interviews lean toward GPU and…](https://www.techinterview.org/companies/fal-ai-interview-guide/)[F**Fireworks AI**related dataOn our AI-Startup Interview Difficulty Index, Fireworks AI ranks 4th of 32 (4.0 out…](https://www.techinterview.org/companies/fireworks-ai/)[G**Genspark**Genspark is a fast-growing Palo Alto AI startup building an agent (its "Super Agent") that…](https://www.techinterview.org/companies/genspark-interview-guide/)[G**Glean**Glean's interview process runs through a recruiter screen, a technical phone screen, and a virtual…](https://www.techinterview.org/companies/glean-interview-guide/)[H**Harvey**Harvey's interview process typically runs through a recruiter screen, one or more role-specific rounds, and…](https://www.techinterview.org/companies/harvey-ai-interview-guide/)[H**Hippocratic AI**Hippocratic AI interviews center on building safe, non-diagnostic patient-facing healthcare LLM agents, so expect rounds…](https://www.techinterview.org/companies/hippocratic-ai-interview-guide/)[H**Hugging Face**A Hugging Face interview tests whether you can build and ship real work with the…](https://www.techinterview.org/companies/hugging-face-interview-guide/)[I**Infinitus Systems**Infinitus Systems builds voice AI agents that phone insurance payers, pharmacies, and providers on behalf…](https://www.techinterview.org/companies/infinitus-interview-guide/)[L**LangChain**LangChain interviews test whether you can build production LLM applications rather than just call a…](https://www.techinterview.org/companies/langchain/)[L**Legora**Legora is a Stockholm-based startup building collaborative AI for lawyers, and it competes head-on with…](https://www.techinterview.org/companies/legora-interview-guide/)[L**LiveKit**LiveKit hires infrastructure engineers who can reason about realtime media and distributed systems, mostly in…](https://www.techinterview.org/companies/livekit-interview-guide/)[L**Luma AI**Luma AI hires for a generative-video and 3D research lab, so the bar skews toward…](https://www.techinterview.org/companies/luma-ai-interview-guide/)[M**Mach Industries**Mach Industries runs a short, sometimes messy loop: a recruiter call, one to three technical…](https://www.techinterview.org/companies/mach-industries-interview-guide/)[M**Magic.dev**Magic (magic.dev) is a small, secretive AI lab building frontier long-context code models, backed by…](https://www.techinterview.org/companies/magic-dev-interview-guide/)[M**Mercor**Mercor screens engineers on practical problem-solving over algorithm trivia: expect a recruiter screen, one or…](https://www.techinterview.org/companies/mercor-interview-guide/)[M**Mistral AI**related dataOn our AI-Startup Interview Difficulty Index, Mistral AI ranks 5th of 32 (3.8 out…](https://www.techinterview.org/companies/mistral/)[M**Modal**Modal interviews test how well you reason about the systems the company actually builds —…](https://www.techinterview.org/companies/modal/)[M**MotherDuck**MotherDuck interviews test your grasp of DuckDB-based analytics and its hybrid execution model, which splits…](https://www.techinterview.org/companies/motherduck/)[O**OpenAI**related dataOn our AI-Startup Interview Difficulty Index, OpenAI ranks 2nd of 32 (4.2 out of…](https://www.techinterview.org/companies/openai/)[O**OpenEvidence**OpenEvidence is a Miami-based clinical AI search engine that answers physicians' questions with citations from…](https://www.techinterview.org/companies/openevidence-interview-guide/)[O**OpenRouter**OpenRouter runs a unified API that routes requests across hundreds of models and providers, so…](https://www.techinterview.org/companies/openrouter-interview-guide/)[O**Oxide Computer**Oxide does not run a normal interview loop: you submit written "materials" (answers to eight…](https://www.techinterview.org/companies/oxide-computer-interview-guide/)[P**Parallel**Parallel Web Systems builds web research and search APIs made for AI agents rather than…](https://www.techinterview.org/companies/parallel-interview-guide/)[P**Perplexity**Perplexity's interview process weighs product judgment as heavily as engineering, so you'll be evaluated on…](https://www.techinterview.org/companies/perplexity/)[P**Physical Intelligence**Physical Intelligence (PI, or "π") builds foundation models for robots, so its bar centers on…](https://www.techinterview.org/companies/physical-intelligence-interview-guide/)[P**Pika Labs**A Pika Labs interview centers on your ability to build and reason about generative video…](https://www.techinterview.org/companies/pika-labs-interview-guide/)[P**Poolside**Poolside interviews skip the LeetCode gauntlet and test whether you can reason about training and…](https://www.techinterview.org/companies/poolside-interview-guide/)[P**Prime Intellect**Prime Intellect builds globally distributed AI training and a GPU compute marketplace, and it hires…](https://www.techinterview.org/companies/prime-intellect-interview-guide/)[R**Reflection AI**Reflection AI is a frontier lab founded by ex-Google DeepMind researchers building autonomous coding agents…](https://www.techinterview.org/companies/reflection-ai-interview-guide/)[R**Reka AI**Reka AI interviews focus on your ability to build and reason about multimodal models that…](https://www.techinterview.org/companies/reka-ai-interview-guide/)[R**Replit**Replit interviews center on practical, ship-it engineering: expect to write and run real code in…](https://www.techinterview.org/companies/replit-interview-guide/)[R**Rillet**Rillet runs a compact, practical loop: recruiter screen, a hiring-manager conversation about backend systems you…](https://www.techinterview.org/companies/rillet-interview-guide/)[R**Rogo**Rogo hires engineers to build an AI analyst for finance, so expect a recruiter screen,…](https://www.techinterview.org/companies/rogo-interview-guide/)[R**Runway**Runway's interview process typically moves through a recruiter screen, one or two role-specific technical or…](https://www.techinterview.org/companies/runway-interview-guide/)[S**Sakana AI**Interviewing at Sakana AI means showing you can do original AI research rather than just…](https://www.techinterview.org/companies/sakana-ai-interview-guide/)[S**Saronic**Saronic runs a fairly standard startup loop: recruiter screen, hiring manager call, then technical rounds…](https://www.techinterview.org/companies/saronic-interview-guide/)[S**Scale AI**Scale AI's 2026 interview process focuses on three areas: designing LLM evaluation frameworks, building RLHF…](https://www.techinterview.org/companies/scale-ai/)[S**Shield AI**Shield AI runs a fairly classic five-stage loop: recruiter screen, hiring manager call, a C++-heavy…](https://www.techinterview.org/companies/shield-ai-interview-guide/)[S**Sierra**Sierra's interview process focuses on how well you can design, build, and reason about production…](https://www.techinterview.org/companies/sierra-interview-guide/)[S**Skild AI**Skild AI builds a general-purpose robotics foundation model (the "Skild Brain") meant to run on…](https://www.techinterview.org/companies/skild-ai-interview-guide/)[S**Stability AI**A Stability AI interview for generative image roles tests how deeply you understand diffusion models…](https://www.techinterview.org/companies/stability-ai-interview-guide/)[S**Suno**A Suno interview tests whether you can design and ship generative audio systems, pairing standard…](https://www.techinterview.org/companies/suno-interview-guide/)[T**Taktile**Taktile is a Berlin- and New York-based decision platform for financial services that lets risk…](https://www.techinterview.org/companies/taktile-interview-guide/)[T**Tempus AI**The Tempus AI interview process moves from a recruiter screen through a technical or role-specific…](https://www.techinterview.org/companies/tempus-ai-interview-guide/)[T**Tennr**Tennr is a New York healthcare-automation company that reads faxed referrals and orders with its…](https://www.techinterview.org/companies/tennr-interview-guide/)[T**TensorWave**TensorWave runs an AI cloud built on AMD Instinct GPUs, such as the MI300X and…](https://www.techinterview.org/companies/tensorwave-interview-guide/)[T**The Browser Company**The Browser Company weights product taste, design sense, and craft as heavily as raw engineering,…](https://www.techinterview.org/companies/browser-company-interview-guide/)[T**Tilde**Tilde (Tilde Research) is a tiny, Khosla-backed AI interpretability lab in Palo Alto hiring ML…](https://www.techinterview.org/companies/tilde-interview-guide/)[T**Together AI**Together AI interviews focus on the engineering behind fast, cheap inference for open models, so…](https://www.techinterview.org/companies/together-ai/)[T**Turbopuffer**Turbopuffer is a small, profitable, hard-core database company (Rust, object storage, search engines) that hires…](https://www.techinterview.org/companies/turbopuffer-interview-guide/)[T**Twelve Labs**Twelve Labs builds video understanding foundation models sold as an API: Marengo for video embeddings…](https://www.techinterview.org/companies/twelve-labs-interview-guide/)[X**xAI**related dataOn our AI-Startup Interview Difficulty Index, xAI ranks 1st of 32 (4.4 out of…](https://www.techinterview.org/companies/xai/)[X**XBOW**XBOW builds an autonomous AI system that finds and exploits software vulnerabilities, and in 2025…](https://www.techinterview.org/companies/xbow-interview-guide/)[Y**Yugabyte**Yugabyte runs a recruiter screen, a hiring manager or tech screen, then four to five…](https://www.techinterview.org/companies/yugabyte-interview-guide/)