The Curious Engineer

All blogs / Compilers

Stephen Diehl

stephendiehl.com

Personal blog of Stephen Diehl - Software engineer writing about technology, programming, and the future

CompilersAI / MLFunctional / PL theoryEssays / personal

’15’20’25
Posts
136 from 21 Apr 2013 to 11 Oct 2026
Posting
monthly (23 in the last year)
Kind
personal blog
Feed
https://www.stephendiehl.com/feed.xml

2026 20

  1. MLIR Part 14 - GPT-2 on the GPU
  2. We Live in the Dependently Typed Future Now
  3. No, Transformers Won't End the Human Race lol
  4. The Internet Is Kind of a Predatory Cesspit Now
  5. Prism: An Impure Functional Language With Typed Effects
  6. MLIR Part 13 - Running GPT-2
  7. MLIR Part 12 - Fusion, Bufferization and Execution
  8. Crypto in 2026: Oh, This is the Bad Place
  9. Book Review: On the Calculation of Volume
  10. MLIR Part 11 - Matrix Multiplication and Attention
  11. Book Review: What We Can Know
  12. MLIR Part 10 - Compiling Transformer Kernels
  13. Book Review: There Is No Antimemetics Division
  14. Book Review: Piranesi
  15. MLIR Part 9 - A Python Frontend for MLIR
  16. Optimal Caverna Gameplay via Formal Methods
  17. Can Opus 4.6 do Category Theory in Lean?
  18. Book Review: When We Cease to Understand the World
  19. MLIR Part 8 - GPU Compilation with MLIR
  20. Hypothetical Divine Signatures

2025 42

  1. MLIR Part 7 - Transformers
  2. MLIR Part 6 - Specializing Python with E-graphs
  3. MLIR Part 5 - Neural Networks
  4. MLIR Part 4 - Linear Algebra in MLIR
  5. MLIR Part 3 - Affine Dialect and OpenMP
  6. Typechecker Zoo
  7. The Stochastic Code Monkey Theorem
  8. Christian Nationalism Is A Dangerous Ideology
  9. It Would Be Good if the AI Bubble Burst
  10. Every Democrat Who Enabled Trump's Crypto Corruption
  11. Only Solution to the Attention Economy is to Opt Out
  12. Fast Tensor Canonicalization in Rust
  13. MLIR Part 2 - Memory in MLIR
  14. Fine-tuning With Tool Calling
  15. Program Synthesis: The λ in the Machine
  16. The Asbestos Doctrine: Why Crypto Must Be Contained, Not Condoned
  17. What is AI market actually worth?
  18. Book Review: Tomorrow, and Tomorrow, and Tomorrow
  19. The Future of Maths May Be Deeply Weird
  20. Does the Star Trek Computer Run on COBOL?
  21. Interfacing MCP with Combinatorial, Convex, and SMT Solvers
  22. The Trump Crypto Orgy of Corruption
  23. Unnecessary and Unstable: Why Stablecoins are Bad
  24. Remote MCP Servers
  25. Adventures in Symbolic Algebra with Model Context Protocol
  26. Using CUDA Deep Neural Network (cuDNN) in Python
  27. The Technofascist Mind: A Guide to Its Psychology and Philosophy
  28. The Kardashev-Marx Scale
  29. AI Slopocalypse 2027
  30. Attention Wasn't All We Needed
  31. MLIR Part 1 - Introduction to MLIR
  32. How Democrats Failed on Crypto
  33. MLIR Part 0 - Installing MLIR
  34. SGLang on AMD MI300X
  35. How to Make a Universe
  36. The FlashAttention CUDA Kernel Line by Line
  37. Training with GRPOTrainer
  38. The Case Against Crypto in 2025
  39. Tokenization is a Meaningless Buzzword
  40. Worse Than a Scam: How Crypto Fueled American Fascism
  41. Enterprise Software in 2025
  42. The Post-literate Economy

2024 34

  1. Sci-fi Hidden Gems
  2. The Sybaritic Economy
  3. Process Reward Models
  4. Digital Dollarization: The Looming Threat to Monetary Sovereignty
  5. Introduction to Fine-tuning Large Language Models
  6. Setting up a Nvidia GH200 for Development
  7. Hymba : Mamba × Transformers
  8. Dependent Types in 200 Lines of Python
  9. Setting up PyTorch for OCaml
  10. New Rank 29 Elliptic Curve
  11. Tiny GraphRAG (Part 2)
  12. Tiny GraphRAG (Part 1)
  13. The Limitations of RAG
  14. The Legacy of Web3
  15. AGI Discourse as Kayfabe
  16. Types of Types: Common → Exotic
  17. Monte Carlo Tree Search
  18. Fine-tuning with ORPO and Unsloth
  19. Pi from String Amplitudes
  20. Diaconescu's Theorem
  21. A Rapid Tutorial on Unsloth
  22. Basic Formal Ontology
  23. The AI Backlash is to be Expected
  24. Volatility Surface
  25. The Dark Eschatology of Crypto Ideologues
  26. Silicon Valley's (totally not a cult) Ideologies
  27. Slippage Modelling
  28. Portfolio Rebalancing
  29. Using SMT Solvers to Manage JIRA Tickets
  30. Word Calculators are Useful, but they're not Smart
  31. Enough Numbers to Build a Universe
  32. Cooking Logics with Soufflé
  33. E-Graphs in Rust
  34. Portfolio Theory with CVXOPT

2023 8

  1. Book Review: The Crisis of Democratic Capitalism
  2. The Rise of Anti-Liberalism in Tech
  3. Why I Left Twitter
  4. Book Review: Capitalist Realism
  5. Best Sci-Fi Books of the 21st Century
  6. Going Deep on DeepSpeed
  7. GPT-2 in One Function
  8. Deconstructing the Worldview of Peter Thiel

2022 7

  1. Crypto is Financial Asbestos
  2. Linking CUDA Kernels into Python
  3. The Public Policy Case for Banning Crypto
  4. Simulating Qubits with Python (Classically)
  5. Book Review: Klara and the Sun
  6. The Token Disconnect
  7. Why Democrats Should Oppose Crypto

2021 4

  1. The Case Against Crypto
  2. Enterprise Eldritch Horrors
  3. Gamestop, Bitcoin and the Commoditization of Populist Rage
  4. Notes on Trekonomics

2019 2

  1. Proving Trivial Theorems in Lean
  2. Listicle of Buddhist Lists

2018 4

  1. What are Elementary Particles?
  2. Integral Zoo
  3. Derivative Zoo
  4. Using Google OR-Tools to do Answer Set Programming

2017 1

  1. The Y-Combinator in Python

2016 11

  1. Dive into GHC: Targeting Core
  2. Dive into GHC: Intermediate Forms
  3. Dive into GHC: Pipeline
  4. Building a Better Custom Haskell Prelude
  5. Exotic Programming Ideas, Part 2 (Term Rewriting)
  6. Exotic Programming Ideas, Part 4 (Datalog)
  7. Cooking Classes with Datatype Generic Programming
  8. Exotic Programming Ideas, Part 3 (Effect Systems)
  9. Exotic Programming Ideas, Part 1 (Module Systems)
  10. From Newton to Einstein and Back
  11. Monads to Machine Code

2013 3

  1. From Adjunctions to Monads
  2. Monads Made Difficult
  3. Space-time Algebra in Python