✍️ 필사 모드: The Developer's Podcast Curation 2026 — From Latent Space to Changelog to SE Daily, the Commute-Sized Signal

Prologue — When 30 minutes a day becomes a year

This post is the third companion to The Developer's Bookshelf 2026 and The Developer's Video and Course Curation 2026, posted the same day. If books grow the skeleton of your judgment and video puts muscle on it, podcasts are the blood vessels running through that muscle. Commute, walk, dishes, drive — the hours when your hands are tied and your eyes are elsewhere — that's the time podcasts convert into the industry's pulse landing inside your head.

Podcasts have three strengths that books and videos don't:

• The time is free. A 30-minute commute, five days a week, fifty weeks a year — that's 125 hours. Enough to finish a serious book, except you're listening sideways.
• The conversational context comes attached. Books give you conclusions; videos give you demos; podcasts give you how someone arrived at their conclusion, in their own voice. Where they got stuck, what they regret deciding, the tone that writing struggles to capture — it's all alive.
• You learn names. A year of one podcast and 30 protagonists in that field are in your head. Their tweets, blogs, papers become the entry points to your next year of learning.

But the traps are equally clear:

• Half-attention drains. A serious one-hour conversation, listened to once, leaves 30% in your head. Twice gets you to 60%. Three times, with notes, gets you to 90%. Good episodes are listened to three times.
• Ads bleed into the content. Every 2026 podcast has sponsors. Ad reads are 30–60 seconds and clearly marked, but the "this episode is brought to you by X" tone sometimes seeps into the main conversation — especially when the guest's company is the topic. Remember that hard questions about a sponsor company get softer.
• Queue inflation. Stuffing 7–10 podcasts into a weekly queue creates guilt, not learning. Keep the queue to 3–5 and triage the rest quarterly.

This post lists the podcasts I've personally listened to for at least a quarter, plus a handful that colleagues recommended twice or more in the past year. For each:

• What signal you get — the one thing you can't get elsewhere
• Who it fits — junior, mid, or senior
• Where to start — 1–3 entry episodes
• Frequency — weekly, biweekly, daily, sporadic

Then I group by category (AI/ML engineering, engineering culture, deep technical, industry strategy, Korean developer podcasts) and close with a listening routine, anti-patterns, and what's next.

Podcasts are a tool for letting information flow through your head, not stick to it. Eighty percent of every episode you listen to should be forgotten without guilt. But the 20% that lingers — I know this person, I know this company, I know both sides of this debate — that 20% is what separates senior from junior over a decade.

---

The category-by-podcast matrix

Fit is suggestive, not absolute. Pick one or two whose tone matches yours, listen for a quarter, then prune. Don't seed the queue with ten on day one.

---

Chapter 1 · AI/ML engineering — Latent Space and friends

The biggest shift of 2024–2026 is that "AI engineer" hardened into a job title. Distinct from ML engineer, distinct from backend. They don't train models — they build systems that use them. Eval, tool use, memory, multi-agent, MCP — this new vocabulary is absorbed fastest through podcasts.

Latent Space — the canon of AI engineering

• Hosts — Swyx (Shawn Wang), Alessio Fanelli
• Cadence — 1–2 episodes per week, 60–120 min
• What signal you get — In 2023 Swyx coined the title "AI Engineer" and then spent the next year building its tooling, vocabulary, and conferences. Guests are builders from OpenAI, Anthropic, Cursor, Vercel, LangChain, Modal, E2B. The canonical "how we're building this right now" source. It's also the official media of AI Engineer Summit and AI Engineer Conference.
• Fit — Anyone shipping software on top of an LLM. Especially RAG, agents, eval, tool use.
• Where to start
  1. The Andrej Karpathy interview — the philosophy of learning in the transformer era
  2. Aman Sanger (Cursor) — the inside of an AI IDE
  3. Mike Krieger / Claude team — the design intent behind Claude and MCP
• Trap — Episodes run long. Listen at 1.5x and pull up the transcript when slang gets thick. Swyx himself sometimes out-talks the guest, which divides listeners.

No Priors — frontier-lab decisions, from the capital side

• Hosts — Sarah Guo (Conviction VC), Elad Gil (angel investor)
• Cadence — weekly, 45–70 min
• What signal you get — If Latent Space is the builder view, No Priors is the capital view. CEOs and CTOs of OpenAI, Anthropic, DeepMind, Mistral, Perplexity are regulars, along with frontier researchers in models, robotics, and biology. Both hosts are first-generation AI investors, so the questions are sharp.
• Fit — Senior engineers and founders who want to read AI company strategy.
• Where to start
  1. The annual Sam Altman episode
  2. The Dario Amodei episode — Anthropic's safety and scaling thinking
  3. Andrej Karpathy "Software Is Changing (Again)" — the podcast version of that famous talk

Dwarkesh Podcast — one topic, deeply

• Host — Dwarkesh Patel
• Cadence — biweekly, 90–180 min
• What signal you get — One guest, one topic, deep. Prep volume is abnormal: Dwarkesh reads every paper, interview, and tweet from the guest before recording. The result is double the density of most podcasts. Guest list is top-tier: Demis Hassabis, Dario Amodei, Sholto Douglas and Trenton Bricken (Anthropic researchers), Sarah C. M. Paine (historian), Mark Zuckerberg, Tony Blair.
• Fit — ML engineers and researchers who want inside the lab. Historians and policy episodes are bonus breadth.
• Where to start
  1. Sholto Douglas and Trenton Bricken on Claude reasoning
  2. Dario Amodei
  3. Demis Hassabis
• Trap — They're long. Genuinely long. Drop the idea of finishing one on the commute; split a single episode across two or three sessions.

Practical AI — code and tools, not vibes

• Hosts — Daniel Whitenack, Chris Benson
• Cadence — weekly, 40–60 min
• What signal you get — Changelog network's AI podcast. Closer to applied code and tooling than to builder interviews. LangChain, LlamaIndex, DuckDB, Weights and Biases are regular topics. Both hosts are practicing data scientists, so the questions are pragmatic.
• Fit — ML engineers, MLOps engineers, data engineers.
• Where to start
  1. The RAG series — five or six episodes in a row
  2. Local LLMs series — llama.cpp, Ollama era
  3. Evaluation series — LLM eval tooling

---

Chapter 2 · Engineering culture and community — Changelog and its universe

Changelog Media started in 2009 as an OSS and developer community outlet. Beyond the flagship podcast it runs JS Party, Practical AI, Go Time, Brain Science, Founders Talk — about a dozen shows. Of any single company's podcast network, it carries the most consistent tone.

The Changelog — the OSS maintainer's living room

• Hosts — Adam Stacoviak, Jerod Santo
• Cadence — weekly, 60–90 min
• What signal you get — Open-source project maintainers and founders as guests. Linus Torvalds, Brian Kernighan, Guido van Rossum, Yukihiro Matsumoto (Matz) have all been on, and today's rising OSS maintainers get the same treatment. The two hosts have worked together for almost twenty years, so the joke timing is good.
• Fit — Anyone who contributes to or uses open source. Which is almost everyone.
• Where to start
  1. The Linus Torvalds episode — 30 years of Linux and git
  2. Guido van Rossum — Python and Mojo
  3. DHH — Rails and SaaS's second spring
• Sibling shows — Changelog Interviews (this one), Changelog and Friends (lighter roundtable), Ship It! (infra and DevOps), Practical AI, JS Party, Go Time.

JS Party — the JS roundtable

• Hosts — A rotating Changelog cast (Amelia Wattenberger, Jerod Santo, Amal Hussein, Nick Nisi, etc.)
• Cadence — weekly, 50–80 min
• What signal you get — Not 1:1 interviews — a 3 to 5 person roundtable on one topic. React, Vue, Solid, Astro, Bun, Deno maintainers sit alongside users. Disagreements are common, and the disagreements are the learning.
• Fit — Frontend, full-stack, Node, Deno developers.
• Where to start
  1. The annual "State of JS" episodes
  2. Ryan Dahl (Deno) episode
  3. Jarred Sumner (Bun) episode

CoRecursive — the story behind the code

• Host — Adam Gordon Bell
• Cadence — monthly to twice-monthly, 50–80 min
• What signal you get — A single project's or single person's story told as a documentary. Not just an interview — there's editing and music and arc, 30 to 80 minutes of audio doc. Episodes like "I was a Coder. So I Fired Myself," "The Mining Pool," "The Coding Career Handbook" feel like watching a video documentary while you walk.
• Fit — Anyone curious about the people and culture behind the code.
• Where to start
  1. "Hard Drive Decade" — the rise and fall of a hard-disk company
  2. "Lord of the Io_Uring" — interview with the Linux io_uring creator
  3. "The Box that Built the Modern Internet" — the Cisco router story

Software Unscripted — languages, runtimes, systems

• Host — Richard Feldman
• Cadence — biweekly, 60–90 min
• What signal you get — Roc language creator Richard Feldman hosts. Guests are mostly language designers and runtime engineers. Deep conversations from the Rust, Zig, Roc, OCaml, Haskell, Gleam camps.
• Fit — Anyone interested in language and compiler implementation.
• Where to start
  1. Andrew Kelley (Zig)
  2. Niko Matsakis (Rust)
  3. The Jane Street OCaml team

---

Chapter 3 · Deep technical and senior engineering — SE Daily and Pragmatic Engineer

Software Engineering Daily — almost every day, one hour

• Hosts — A rotating cast (Jeff Meyerson founded; today Jordi Mon Companys and 5–6 others)
• Cadence — near-daily, 50–70 min
• What signal you get — One company, one tool, one domain, one hour at a time. The breadth is massive: databases, search, system design, blockchain, observability, security, embedded — niches no other podcast covers. The volume means you have to learn how to pick.
• Fit — Anyone with wide curiosity. Especially when you want a one-hour primer on a domain outside your own.
• Where to start
  1. Search the archive by your favorite keyword and pick the three most recent
  2. Annual "Top 10 of [year]" lists
• Trap — Founder Jeff Meyerson died in 2021. The show's tone wobbled for a year and stabilized after 2024 with a new host pool. Guest accents are varied, so the transcript helps.

The Pragmatic Engineer Podcast — Big Tech internals

• Host — Gergely Orosz
• Cadence — biweekly, 60–90 min
• What signal you get — Audio companion to the same-named newsletter (700K+ subscribers). The best public window into what Big Tech is actually doing internally. Senior engineers and managers from Meta, Stripe, Uber, Shopify, Datadog. First-person on "why this company made this decision."
• Fit — Mid and senior engineers, engineering managers, anyone on a path to senior.
• Where to start
  1. "Inside Stripe Engineering" — conversation with the Stripe CTO
  2. "On-call at Big Tech" — on-call culture
  3. "Engineering Career Frameworks" — IC vs manager tracks
• Trap — Gergely spends ~30 hours a week on the newsletter. The podcast is somewhat secondary, so the biweekly cadence wobbles. Pair it with the newsletter for full effect.

Oxide and Friends — systems, hardware, culture

• Hosts — Bryan Cantrill, Adam Leventhal
• Cadence — weekly, 60–120 min
• What signal you get — The co-founders of Oxide Computer Company. Thirty years working together at Sun Microsystems and Joyent. Systems engineering and culture roll together. Their fellow Oxide engineers frequently join as guests.
• Fit — Systems engineers, backend folks close to hardware, and any senior who cares about engineering culture.
• Where to start
  1. The "Soul of a New Machine" series — not a book review, but what the book teaches
  2. "Predictions for 2026" — their annual new-year episode
  3. Open-source strategy episodes — OSS and capital

Software Engineering Radio — bridging academia and industry

• Hosts — IEEE Software editors (Robert Blumen, Jeremy Jung, rotating)
• Cadence — biweekly, 45–75 min
• What signal you get — Official podcast of IEEE Software. Guests are book authors, academics, senior engineers. Tone is more serious than most podcasts, and small talk is minimal. Feels like compressing one technical book into one hour.
• Fit — Serious learners; people for whom listening to a whole English book on tape feels too heavy.
• Where to start
  1. Martin Kleppmann — DDIA episode
  2. Camille Fournier — Manager's Path
  3. Gregor Hohpe — Enterprise Integration Patterns

---

Chapter 4 · Industry strategy and business — widening your field of view

Acquired — one company, told whole

• Hosts — Ben Gilbert, David Rosenthal
• Cadence — biweekly, 180–300 min (each is 3–5 hours)
• What signal you get — One company, from founding to today. Episodes typically run 3–5 hours; guest episodes are shorter. They've done NVIDIA, Microsoft, Costco, LVMH, TSMC, Standard Oil, Berkshire Hathaway. The NVIDIA trilogy is widely cited as some of the best business content of 2024.
• Fit — Seniors and founders who want to widen their view. Anyone setting their company's strategy.
• Where to start
  1. NVIDIA Parts I, II, III — built over three years from 2022–2024
  2. TSMC — the semiconductor supply chain
  3. Microsoft — the Satya-era reinvention
• Trap — They're long. One episode spreads over two weeks of commutes. But two episodes a year is enough to get the inside of one industry into your head.

Decoder with Nilay Patel — top-tier CEO and policy interviews

• Host — Nilay Patel (editor-in-chief, The Verge)
• Cadence — 2–3 per week, 60–90 min
• What signal you get — CEOs, policymakers, media chiefs. Nilay is a former lawyer, so the questions are sharp and he re-asks when a guest dodges. Sundar Pichai, Mark Zuckerberg, Sam Altman, Jensen Huang, Lina Khan — alongside small startup CEOs. Paired with The Vergecast (Fridays).
• Fit — Anyone tracking industry, policy, or platform power. Senior engineers, managers, founders.
• Where to start
  1. The post-2024 Sam Altman episodes
  2. Lina Khan (former FTC chair)
  3. Jensen Huang

Stratechery Podcast — daily 30 minutes of industry analysis

• Host — Ben Thompson (sometimes with guests)
• Cadence — weekdays, 30–60 min
• What signal you get — Not the audio version of the newsletter — separate content. Ben Thompson has applied the same analytical frame (Aggregation Theory, 5-forces variants) to industries since 2013. The value is consistency of thinking. Sister shows Sharp Tech (developer-tech angle), Sharp China (China tech industry), Greatest of All Talk (NBA-as-industry metaphor).
• Fit — Senior engineers and founders building strategic thinking. The daily 30 minutes is manageable.
• Where to start
  1. The "Aggregation Theory" introductory episode (the web essay version is also free)
  2. The NVIDIA, Microsoft, Apple analysis arcs
• Trap — Most content is behind a Stratechery subscription. Sample the free preview episodes first.

20VC — VCs and founders, three times a week

• Host — Harry Stebbings
• Cadence — 3–5 per week, 40–70 min
• What signal you get — Harry started this at 20 in 2015 and now runs his own fund. Three to five episodes a week — top VCs, top founders, top CEOs. The depth of the guest roster is unmatched, though some critics find the host's questions thin.
• Fit — Anyone interested in VC, founding, or strategy.
• Where to start
  1. Doug Leone (Sequoia)
  2. Brian Chesky (Airbnb)
  3. Jensen Huang

Lenny's Podcast — PM, growth, organization

• Host — Lenny Rachitsky
• Cadence — weekly, 60–90 min
• What signal you get — Audio companion to the giant PM newsletter (800K+ subscribers). Guests are PMs, designers, growth and org leaders. Shreyas Doshi, April Dunford, Marty Cagan, Julie Zhuo. Officially a PM podcast, but top-tier for senior engineers and tech leads too.
• Fit — Mid and senior engineers, engineering managers, any IC who works with PMs.
• Where to start
  1. Shreyas Doshi — "What separates great PMs from the rest"
  2. April Dunford — "Obviously Awesome" positioning
  3. Marty Cagan — the "Empowered" series

---

Chapter 5 · Broad conversations and culture — Lex and friends

Lex Fridman Podcast — long, deep interviews across everything

• Host — Lex Fridman
• Cadence — 1–2 per week, 120–300 min
• What signal you get — AI researchers, politicians, physicists, founders, athletes, chefs. The widest guest range in podcasting. Episodes are typically 3–5 hours, Joe-Rogan-style long-form. Guests get asked questions they rarely face elsewhere — life, death, love, faith.
• Fit — Anyone with broad curiosity. Speed up and jump around freely.
• Where to start
  1. The annual Andrej Karpathy episode
  2. Demis Hassabis
  3. Stephen Wolfram
• Trap — The host's earnest tone divides listeners. If you bounce off the first 30 minutes, skip that episode entirely.

a16z Podcast — portfolio and adjacent

• Hosts — Multiple (Sonal Chokshi and a16z partners)
• Cadence — 2–3 per week, 30–60 min
• What signal you get — Andreessen Horowitz's official network. Plenty of portfolio companies appear, but the network splits by sector (a16z AI, a16z Bio Eats World, etc.), so you can filter.
• Fit — Anyone at the venture-tech intersection.
• Trap — Soft questions for portfolio guests. Keep that in mind.

The TWIML AI Podcast — bridging ML academia and industry

• Host — Sam Charrington
• Cadence — 1–2 per week, 45–75 min
• What signal you get — "This Week in Machine Learning AI." ML interviews since 2016. Covers both academic researchers and industry engineers. Like Dwarkesh, Sam reads the guest's papers and code before the interview.
• Fit — ML engineers and researchers.
• Where to start
  1. The NeurIPS and ICML conference special episodes
  2. Episodes timed with major model releases — the model authors

---

Chapter 6 · Korean developer podcasts — the language-of-home retrospective

Listening only to English podcasts costs you something: time spent reflecting on the same market in the same language. Korean dev podcasts have thinned out, but a few continue, and the tone is more serious for it.

NaPDa (나는프로그래머다, "I Am a Programmer")

• Hosts — Baekjun Lim, Sangmin Lee, Jongwoo Lee, Jaegyung Song (rotating)
• Cadence — Sporadic (effectively dormant since 2020; archive of 200+ episodes alive)
• What signal you get — The most influential Korean developer podcast of 2014–2019. Senior Korean engineers, led by writer Baekjun Lim, talking honestly about career, technology, industry. New episodes are rare, but the archive of 200+ episodes is enough on its own.
• Fit — Korean senior engineers, managers, and anyone wanting context on Korea's IT industry in the 2010s.
• Where to start
  1. "Korean Developer Writing" series
  2. "What Is a Senior Developer" roundtable
  3. The 30-year retrospective on Korean IT

Developer Radio (개발자 라디오)

• Hosts — Yousung Kang, Jio Shin and others (host roster rotates by season)
• Cadence — Biweekly to monthly, 50–80 min
• What signal you get — Korean backend and full-stack engineers talking through daily technical decisions. The tone is close to the ground: less "this is how Coupang or Toss did it" and more "this is the problem our small team had, and this is how we solved it." Low on humblebrag, which is the strength.
• Fit — Korean mid-level developers. Most useful in the 1–5 year range.
• Where to start
  1. The first 1–3 episodes of the current season — to check the host tone
  2. The side-project series
  3. The job-change and compensation series

Yozm IT Podcast (요즘IT 팟캐스트)

• Host — Yozm IT editorial staff
• Cadence — weekly, 30–50 min
• What signal you get — Audio counterpart of Yozm IT (web media). Short Korean takes on companies, people, trends. The depth is shallow, but it's the fastest way to track week-by-week change in Korea's IT industry.
• Fit — Anyone watching Korea's IT industry. Useful when you're thinking about a job change or founding something.
• Where to start — Listen to the most recent 4–5 episodes back-to-back. The recurring patterns become visible.

Indie Korean dev podcasts

• Since 2020, several new indie Korean developer podcasts have appeared. Hosts, cadence, and lineups change frequently, so check the list quarterly. Search terms: "Korean developer podcast," "한국 개발자 팟캐스트," "기술 팟캐스트 추천."
• Korean YouTube video-podcasts effectively fall in the same bucket. Channels like LightHacker, Nomad Coders' interview series, Woowahan Tech Seminar, and Inflearn Live publish video conversations that work well as audio.

---

Chapter 7 · A listening routine — making podcasts add up

Podcasts have two modes: half-attention listening and serious listening. You need both to retain anything after a year.

7-1 · Queue management — the 3·5·1 rule

• 3 weekly podcasts — the ones you listen to every new episode. Examples: Latent Space, Changelog, The Pragmatic Engineer.
• 5 biweekly/sporadic podcasts — pick by title. Examples: Software Engineering Daily, Decoder, Acquired, No Priors, Dwarkesh.
• 1 deep podcast — one quarter-spanning long-form. One Acquired arc, one Dwarkesh interview, one Lex Fridman with a particular guest. Serious weekend listening.

Total: about 9–10 episodes a week, average 60 minutes each, ~9–10 hours. Easily absorbed across commute, exercise, and chores.

7-2 · The listening stages — let it flow, then go back

1. First pass — flow listening. Once through at 1.2–1.5x. No transcript.
2. Mark candidates. When something hits, drop a one-line voice memo: "re-listen to this part."
3. Second pass — serious re-listen. Only the marked sections, at 1x, with the transcript open. Usually 10–20 minutes.
4. Write a paragraph. 5–10 lines in your notes — enough that a month later you could explain the episode to someone who hasn't heard it.

You only need 1–2 serious-listen episodes a week. The rest is fine as flow listening.

7-3 · Transcripts — the new 2026 tool

Since 2024 almost every top-tier podcast publishes transcripts. Apple Podcasts auto-transcribes, Spotify ships transcripts, and most podcasts post them on their own site.

• Open the transcript when names blur past. Missing guest names, tool names, paper titles is the most common loss.
• AI summaries as helpers, not replacements. Snipd, Podwise, Recall turn an episode into a 5–10 minute card. Useful for organizing after a serious listen. Don't mistake "I read the summary" for "I listened."

7-4 · A sample weekly schedule (~9 hours)

• Commute, 30 min x 5 days (2.5 h) — One new Latent Space or Changelog per week.
• Lunch walk, 20 min x 3 days (1 h) — Short Stratechery or Decoder.
• Evening workout, 40 min x 3 days (2 h) — One of Pragmatic Engineer, SE Daily, No Priors.
• One weekend hour — A third of an Acquired or half a Dwarkesh.
• Spare slot — Lex Fridman, Lenny's. Optional queue.

---

Chapter 8 · Anti-patterns — what podcasts can't do

• The "I listened, therefore I learned" trap. Without typing one line of code, almost nothing sticks. Serious listening has to pair with notes.
• The queue-growth trap. A new podcast added every week, queue at 30 by year's end. Episodes pile up unfinished and the guilt accumulates. Triage quarterly; add new shows one at a time.
• The summary-only trap. Snipd and Podwise are aids. If you only read the card, you lose tone, nuance, and the silences in the conversation. That's not learning.
• The 1.8x trap. Trying to clear the commute faster, you bump speed to 1.8x or 2x. Five minutes in, your head hurts. 1.2–1.5x is the practical ceiling.
• The "sponsor read is a signal" trap. The X-brought-to-you-by-Y is an ad. Same for when a guest praises their own company. Capacity to ask hard questions about a sponsor's company goes down — remember.
• The English-only trap. You lose retrospection in your own language and your own market. Mix in a Korean or Japanese show. Mother-tongue listening shapes a different circuit in your head.
• The senior-interview-only trap. A week of Acquired, Lenny's, Pragmatic Engineer fills your head with people ten years ahead of you. Your current work starts to feel ungrounded. Balance with your peers' tone — Latent Space, Changelog, JS Party.

---

Epilogue — Take the earbuds out and go back to the code

The value of podcasts is that they're a tool for spending time twice. The 30-minute commute is going to happen either way. Whether it leaves the industry's pulse in your head depends entirely on queue management.

Learning in 2026 is layered. Books for principles, video for tools, podcasts for pulse, an AI co-tutor for the gaps, your own codebase for application. All five — anything less and the year's learning won't survive the next year. None of them replaces another.

Checklist:

• Three weekly podcasts chosen — neither too few nor too many
• Five biweekly/sporadic queue items pruned at the quarter
• At least one episode seriously re-listened this month
• At least one new name or tool from those episodes added to your notes
• Playback speed not above 1.5x
• At least one mother-tongue podcast in the queue
• Sponsor reads mentally separated from the main conversation

Next post — The Developer's Newsletter Curation 2026. The Pragmatic Engineer, Lenny's, Stratechery, Bytes, JavaScript Weekly, Hacker Newsletter, Import AI, Last Week in AI — top-tier newsletters by category. Newsletters sit between books and podcasts: shorter than a podcast, faster than a book. How to fill that slot.

---

참고 / References

AI/ML engineering

• Latent Space — https://www.latent.space/
• No Priors — https://www.no-priors.com/
• Dwarkesh Podcast — https://www.dwarkeshpatel.com/podcast
• Practical AI — https://changelog.com/practicalai
• The TWIML AI Podcast — https://twimlai.com/podcast/

Engineering culture and OSS

• The Changelog — https://changelog.com/podcast
• JS Party — https://changelog.com/jsparty
• Go Time — https://changelog.com/gotime
• Ship It! — https://changelog.com/shipit
• CoRecursive — https://corecursive.com/
• Software Unscripted — https://softwareunscripted.com/

Deep technical and senior engineering

• Software Engineering Daily — https://softwareengineeringdaily.com/
• The Pragmatic Engineer Podcast — https://newsletter.pragmaticengineer.com/podcast
• Oxide and Friends — https://oxide.computer/podcasts/oxide-and-friends
• Software Engineering Radio — https://se-radio.net/
• The Stack Overflow Podcast — https://stackoverflow.blog/podcast/

Industry strategy and business

• Acquired — https://www.acquired.fm/
• Decoder with Nilay Patel — https://www.theverge.com/decoder-podcast-with-nilay-patel
• The Vergecast — https://www.theverge.com/the-vergecast
• Stratechery Podcast — https://stratechery.com/category/podcasts/
• 20VC — https://www.thetwentyminutevc.com/
• Lenny's Podcast — https://www.lennyspodcast.com/

Broad conversations and culture

• Lex Fridman Podcast — https://lexfridman.com/podcast/
• a16z Podcast — https://a16z.com/podcasts/

Korean developer podcasts

• 나는프로그래머다 (NaPDa) archive — https://www.nadocast.com/
• 개발자 라디오 (Developer Radio) — https://www.developer-radio.com/
• 요즘IT (Yozm IT) — https://yozm.wishket.com/

Must-listen episodes (direct links)

• Latent Space, "Andrej Karpathy" — https://www.latent.space/p/karpathy
• Latent Space, "Sourcegraph + Cursor: AI Coding" — https://www.latent.space/p/ai-coding
• Latent Space, "Anthropic's Mike Krieger" — https://www.latent.space/p/mike-krieger
• Dwarkesh, "Sholto Douglas and Trenton Bricken on Claude" — https://www.dwarkeshpatel.com/p/sholto-trenton-2
• Dwarkesh, "Dario Amodei" — https://www.dwarkeshpatel.com/p/dario-amodei
• No Priors, "Sam Altman" — https://www.no-priors.com/episodes/sam-altman-openai
• Changelog, "Linus Torvalds" — https://changelog.com/podcast/400
• Changelog, "Guido van Rossum" — https://changelog.com/podcast/420
• The Pragmatic Engineer, "Inside Stripe" — https://newsletter.pragmaticengineer.com/p/podcast-stripe
• Acquired, "NVIDIA Parts I, II, III" — https://www.acquired.fm/episodes?q=nvidia
• Lex Fridman, "Andrej Karpathy" — https://lexfridman.com/andrej-karpathy/
• Lenny's Podcast, "Shreyas Doshi" — https://www.lennyspodcast.com/shreyas-doshi/
• Oxide and Friends, "Predictions" — https://oxide.computer/podcasts/oxide-and-friends

Tools

• Snipd (podcast AI summaries) — https://www.snipd.com/
• Podwise (episode summaries) — https://podwise.ai/
• Overcast (iOS player) — https://overcast.fm/
• Pocket Casts (cross-platform) — https://pocketcasts.com/

Companion posts

• The Developer's Bookshelf 2026
• The Developer's Video and Course Curation 2026