Ten years of the "Deep Learning from Scratch" series reaching LLMs is a perspective almost nobody else has, and the build-from-scratch framing is what makes these books age so well.
@whats_ai
-
27B Dense Model Outperforms 397B MoE on Coding Benchmarks
By
–
A 27B dense model beating the 397B-A17B MoE on coding benchmarks is quite cool, and a better signal than the open-source-catches-closed headline it'll get flattened into.
-
Native Charts and Infographics Output for Research Workflows
By
–
Thanks for the write-up; really looking for the native charts and infographics output! Most research workflows the bottleneck is turning the answer into something you can paste into a doc.
-
Automating nightly conflict summaries with idle compute resources
By
–
This is a smart use of idle compute; the piece I'd build in is a nightly summary flagging any conflict resolved with non-trivial changes so the morning review stays short.
-
Engineering Ethics: When Not to Build with LLMs
By
–
We’re entering a phase where “can we build it?” matters less than “should we build it this way?” Energy, cost, and latency are becoming very important societal decisions. LLMs are powerful, but they’re not always the right abstraction. Good (AI) engineering is knowing when not
-
Agentic AI System for Automated Research and Writing Workflow
By
–
Our 2-hour workshop from the AI Engineer conference in London is live now!! @pauliusztin_ , Samridhi from @towards_AI , and I walked through the agentic AI system we built. We covered the deep research and writing workflow we built for automating parts of our work. For me,
-
Who’s Really Stealing From Whom in AI
By
–
Who’s Really Stealing From Whom?
— Louis-François Bouchard 🎥🤖 (@Whats_AI) 21 avril 2026
For the X addicts who don’t leave this app… here’s my latest video 👇 pic.twitter.com/8JVNrxkJW3Who’s Really Stealing From Whom?
For the X addicts who don’t leave this app… here’s my latest video -
Junior Developer Squeeze Elevates Creative Architecture Value
By
–
Junior-SWE squeeze is real, but the creative-architect layer just got more valuable, not less.
-
Token Price Stability Undermined by Tokenizer Drift
By
–
Love that you built the counter, per-token price stability is a lie the moment tokenizer behavior drifts.
-
Harnesses gaining equal recognition with AI models in surveys
By
–
Harnesses finally getting equal billing with models in survey form, this was overdue.