Technically
AI Reference
Your dictionary for AI terms like LLM and RLHF
Company Breakdowns
What technical products actually do and why the companies that make them are valuable
Learning Tracks
In-depth, networked guides to learning specific concepts
Posts Archive
All Technically posts on software concepts since the dawn of time
Terms Universe
The dictionary of software terms you've always wanted

Explore learning tracks

AI, it's not that ComplicatedAnalyzing Software CompaniesBuilding Software ProductsWorking with Data Teams
Loading...
PodcastI'm feeling luckyPricing
Log In
Technically Speaking, hosted by Megan Reynolds, presented by Railway
Episode 3 · Premieres Thursday, November 5

What DeepSeek figured out, and why you shouldn't copy it

For years the recipe for better AI was simple: more data, bigger models. Then DeepSeek's R1 showed models could learn to reason by practicing on problems with checkable answers, and Reinforcement Learning became the way forward.

Follow the show
YouTubeApple PodcastsSpotify
Or use the RSS feed
Julien LaunayCofounder at Adaptive MLSesh NallaVP of Engineering at DatadogMegan ReynoldsHostFounder at infra.nyc

Presented by our friends at Railway, the all-in-one intelligent cloud provider.

Get it in your inbox the day it's out

Every Thursday, with show notes, in the Technically newsletter.

Subscribe
Content
  • All Posts
  • Podcast
  • Learning Tracks
  • AI Reference
  • Companies
  • Terms Universe
Company
  • Pricing
  • Sponsorships
  • Contribute
  • Contact
Connect
SubscribeSubstackYouTubeXLinkedInInstagram
Legal
  • Privacy Policy
  • Terms of Service

© 2026 Technically.