I can't believe I'm saying this but I think typing is going away for most computer operations. I think it'll still be around for deep thought / precise edits like a pen + paper or edits but I've found speaking to your computer is just much faster / less friction on average.

I migrated my markdown blog from C# to Rust: - Latency: 2-8x improvement - Memory: 4x less usage Benchmarking my Markdown Blog in Rust and C# - hamy.xyz/blog/2026-02...

Python + Agents = Fast + Fragile Rust + Agents = Fast + Correct You trade some speed for more reliability.

I thought cross-platform builds for my game would be hard but Raylib and C# / dotnet made it easy to build single-file executables for Linux, Mac, and Windows. I now have automated builds on every push with a version bump. How I Built a Clicker Game with Raylib + C# - hamy.xyz/blog/2026-01...

The biggest risk of fully autonomous AI coding isn't bugs, it's comprehension debt. Your codebase evolves whlie you're not looking and now you don't understand how it works, what tradeoffs were made and why, and lose confidence in the direction to push the project.

The problem (and power) of agents is their nondeterminism. Like a human they need a deterministic runtime/compiler/linter to double check their work. Rust is that deterministic foundation - you get its reliability at the iteration speed of agents.

Rust can approach C in speed, has expressive types, and a solid community growing ~30% a year. The catch: An infamous learning curve and ongoing devx issues. My approach is high-level Rust - get the devx of C# for the cost of 10-20% of Rust's perf ceiling.

I haven't been this excited about a programming language since I found F#. Rust has great types, great performance, and a growing ecosystem. The devx overhead is the only thing holding it back. So naturally I built a boilerplate to try and make up the difference.

With Hypermedia frameworks the backend controls the frontend so you can essentially build with whatever stack you want. Here's a Rust app with Datastar - hamy.xyz/blog/2026-03...

I've written Python, F#, C#, Java, and TypeScript professionally. But with AI writing most of the code I stopped picking langs for my own expertise and started picking them for foundational attributes like types, pefromance, and tooling. That's how I ended up with Rust.

The performance difference between my C# and Rust markdown blog doesn't really matter - sub-ms vs low-ms is imperceptible to users. But the difference in memory usage does, it allows me to run more apps per instance. - Rust baseline: 34 MB - C# baseline: 166 MB

Rust is underrated for agentic engineering: - Agents remove much of the learning curve barrier - Types + Compiler catch many bugs agents (and humans) commonly write - AI is good at Rust and you get high perf systems Move fast with stable infrastructure.

Why High-Level Rust Wins Over Other High-Level Languages for Agentic Engineering - Expressive types + compiler hints - Mainstream lang w long-term support - AI is proficient - Great performance story hamy.xyz/blog/2026-04...

I host my apps in a multi-server configuration for a few reasons: * Learn how to do it, force myself to use infrastructure as code * Improve the existing DB bottleneck * Unlock future scalability w multi-server How I host my apps - hamy.xyz/blog/2025-12...

AI orchestrators can be expensive. Mine cost $90 for ~10 hours of autonomous work, completing 15 small-medium tasks. This is more than I'd want to pay every day and is expensive for a tool but pretty cheap for an engineer.

Agents are fast. It's well worth sacrificing some of it to improve quality. Take some of that saved time to cleanup the code, do the refactor, and improve guardrails. That's how you get higher quality software faster with agents. Move fast with stable infrastructure.

Benchmark breakdowns of what makes AI good at a language: - Training data volume 40% - Syntax simplicity <10% - Compiler errors <10% Rust has ~15% dev usage, ranked top 15 lang. Other langs with expressive types like F#, OCaml, Haskell all <3%. That gap is the whole game for agentic engineering

Programming language non-negotiables in the age of agentic engineering: - Expressive types - Strong ecosystem - libraries, docs - AI good enough to one or few shot features - Decent performance

Why Rust Wins in the Age of AI: - Rust provides a deterministic foundation to build system on - AI agents provide nondeterministic ability to change the system Together you can move fast on a stable foundation. hamy.xyz/blog/2026-04...

Another one. FWIW Claude Code is not currently operational for me.

Want a reactive webapp without a JS framework? Hypermedia allows backend to control the frontend. - Simpler than SPAs - More interactive than MPAs Getting Started with Datastar - Build a Rust + Axum Todo App - hamy.xyz/blog/2026-03...

In the age of AI, I'm finding myself more drawn to terminal-based workflows: - AI CLIs are mainstream - Flip through windows / sessions fast - Portable across machines - High customization / scriptability Why I'm moving to a Linux-based, terminal-focused dev workflow - hamy.xyz/blog/2026-03...

What if you just write Rust like a high-level, garbage collected language? Turns out with a few rules you can basically get all its perf / type benefits with only a 10-20% per hit. - Immutable types, functionalish pipes - Arc + clone liberally - DDD services

AI will change everything. But everything's been changing since the beginning of time. This time will be different but also much the same. The strategy remains: Learn the new tools, adapt to the new environment, and carry on. Will AI replace software engineers? hamy.xyz/blog/2025-12...

Datastar is a lot like combining HTMX + Alpine together in one library. - Server interactivity with SSE - Client reactivity with signals Rust + Datastar - Getting Started with Datastar - Build a Rust + Axum Todo App - hamy.xyz/blog/2026-03...

Why I keep building boilerplates: - Every project needs similar plumbing - Edge cases are easy to miss when bored, rushing, or AI-generating - Fixes patch upstream so every future project starts better

Datastar takes a different approach to Hypermedia than HTMX: - SSE first - HTML fragments + signals in same endpoint - Built-in Reactive signals Getting Started with Datastar - Build a Rust + Axum Todo App - hamy.xyz/blog/2026-03...

C# is currently the closest to S-Tier for me with a 4.5/5. If native unions land well in C# 15, this could be the language I've been looking for. The primary thing missing is expressive types and native unions get it very close.

In 2026.Q1 I participated in a 12-week programming retreat, published my 1,000th blog post, started building with Rust, shipped my first game, played Pokopia, and might've found my second mountain. hamy.xyz/blog/2026-q1...

In Rust, cheap and expensive clones look exactly the same: - cheap_arc.clone() - ~10ns - expensive_string.clone() - 100+ns They look identical in code so one sloppy refactor can silently 10x your clone costs. So I built LightClone to allow the type system to warn you about these.

AI reduces traditional switching costs between programming languages. Your own proficiency is a secondary factor to the AI's. This means you can focus on foundational attributes like types, perf, and tooling with less regard for your own fluency. This is a major factor in pushing me towards Rust.

What I built at Recurse Center: - Applied AI for Engineering - Web Systems with Rust - Networked Games Plus a LOT of writing.

I migrated my markdown blog from C# to Rust and benchmarked them: - Baseline: 34 MB (Rust), 166 MB (C#) - Under load (20k requests, 100 concurrent): 63 MB (Rust), 293 MB (C#) Benchmarking my Markdown Blog in Rust and C# - hamy.xyz/blog/2026-02...

Why I'm moving to a Linux-based, terminal-focused dev workflow - and what it looks like - Ghostty terminal, zellij multiplexer - Fedora distro, Sway wm - neovim + lazygit - Obsidian for notes - Agents, agents everywhere hamy.xyz/blog/2026-03...

I use Ansible to rollout my infrastructure in phases: * Foundations - Hardens OS, connects to Tailscale, installs Docker * Core services - Sets up the special analytics and db roles * App servers - Configures the servers w env vars pointing to the other roles hamy.xyz/blog/2025-12...

Here we go again - another Claude Code outage.

My AI orchestrator isn't very fast nor particularly cheap but it works surprisingly well. It completed 15 tasks in 10 hours fully autonomously for the price of $90 in tokens. More expensive than I'd want to pay every day but far less expensive than a typical software engineer.

The downfall of Claude / Anthropic may not be OpenAI, the US Govt, or increasing costs. It might just be unreliable systems pushing ppl to try alts. Claude is down again (multiple this week) so now I'm trying opencode. status.claude.com

I've never written a book but I've been thinking about it. It seems I could write a minibook each year by compiling blog posts into a book. They would need to be ironed out and edited but seems like it would get 80%. How Good Blogging Habits Yield a Book Each Year - hamy.xyz/blog/2026-01...

What if you could get all of the benefits of Rust (performance, expressive types, solid ecosystem) with none of the pain (devx + onboarding)? With a few rules, you can write it like F#/C#/TS. - Immutable data, functionalish pipes - Arc + clone liberally - DDD services

My current desk setup: * Dual monitors, centered around usage * Standing desk that's mostly seated but standing for meetings * Adjustable height and arms office chair How to setup an ergonomic desk - hamy.xyz/blog/2025-09...

API scaffolding, auth, background jobs, CSS pipelines, and observability are all interesting to build out. But by the 5th or 6th time, it's just overhead. That's why I build boilerplates for myself - build once, harden over time, skip the dev time on the next project.

I built my stats page with no JS: * HTML: SSR rendered with C# + CinderBlockHtml * Chart: Raw SVG, vibe coded hamy.xyz/blog/2026-01...

Why I'm moving from C# to Rust for High-level Apps - hamy.xyz/blog/2026-02... - Expressive types - Great tooling - Strong performance characteristics With similar devx, ecosystems, and community.

The best way to get started with errors as values is by using a Result<Success, Failure> type. * Success - Payload to return on success * Failure - Payload to return on failure This allows each function to declare the exact contract on success and failure. hamy.xyz/blog/2025-07...

Benchmarking my Markdown Blog in Rust and C# - 4x less memory - 4x faster Same logic, same styles, same 1k posts. hamy.xyz/blog/2026-02...

Vimium in Chrome is giving me a lot of joy

AIs are great for first pass code reviews: - Have context of all the code you've written, can do more research - Available 24/7 - Can ask it to do multiple passes - Can run tests, linters, type checkers, etc How I Actually Code with AI - hamy.xyz/blog/2025-10...

My AI orchestrator isn't even that good. - I built it in a few days - It's slow (15 tasks in 10 hours) - It's expensive ($9/h) But if a weekend project can do this, imagine what's coming in the next 6-12 months.

My current mental model for working with autonomous agents is like gardening: - You create the design / scaffolding - You plant seeds (tasks) - You leave, the plants grow (AI working on them) - You come back and tend them - inspecting, refining, redirecting