Mike McQuaid

The wonderful Oliver Whitwam has written a short blog about how we're practicing type-driven development and domain-modelling to create software that is both correct and is communicable via ubiquitous language. We're really making good usage of the Rust newtype and Haskell-esque smart-constructor patterns to leverage types to encode business rules and invariants (until Rust adds refinement types via pattern-types!). If you love #types, #correctness, #domainmodelling, you'll like this blog I'm sure! https://lnkd.in/eqKWXtqS

15

1 Comment

The purpose of this post is to touch base on Ollama and how it can help with both privacy and running a cost efficient pipeline for indie devs. If you haven't heard of ollama, I highly recommend checking their website and read https://lnkd.in/gUvgcRSY. In a nutshell, it allows you to run certain models locally or in a docker container which can help with hosting models on your machine or server. It has a lot more features which I'm still learning about. I've been playing around with qwen3:30b for majority of the time and just recently updated to qwen3.5:35b. Both run smoothly on M4 Max / 48GB but could've run better on 64GB only if I knew better to get that upgrade. I'll write more on the projects that I'm working on in future, but in a nutshell, I used LLM to run a two-phase pipeline: - Filter 1.3million records from jsonl and reduce it to 50K useful records for the app - Run through all 50K records to select most meaningful samples among several examples for each record. In addition, fallback mechanism for any missing data needed

11

6 Comments

Thoughts on Go after using it recently: - not being object orientated is a big plus. One of my least favourite parts of C#/Java - feels right at home coming from a typescript background - concurrency built into the language feels very nice - wish the type system had some more advanced features like discriminated unions - auto garbage collection makes the language very easy to use  - the fact that it compiles to a binary so no runtime is needed is a great advantage - seems like it's a language in between rust and C# which makes it a great tool for a lot of projects - love that go has errors as values and you can see which functions can return an error ahead of time, maybe one day we'll get something like this into typescript ( kinda is with effect.ts ) - interfaces being implicit instead of explicit is interesting and I'm not sure how I feel about it yet

6

I have so much to talk about, but little time to post. If you're stuck on any agentic coding flow issues feel free to send me a message or comment on what you're running into. I might start sessions where I demonstrate how we're approaching agentic development flows for regulated industries. General tips: Files not Agents touching API's - i.e. agent writes a JSON that uploads your data via API - Do a straight copy pass to capture anything you need to be copied 1:1 - Do a second pass where you add the generative descriptions - Have the agent run a script to push the JSON via the API Document all doom loops and save as an MD for future agent reference. Have the agent write a mini postmortem and save it as an MD file for personal/org reference. Write recipes for modular pieces of software that are reused across apps. Include proper architecture and secure usage and leverage this context to build and audit with. Most are probably doing this and more like myself, what other cool harnessing or auto context generation techniques are you all using out there to keep your orgs on track and compliant?

3

While I get this statement, it's important to understand that it's not only about understanding but also about making sure it follows a contract you can trust when you modify it. It's not so much about whether I can or not understand the code, it's about how reliably I can change it. This is where I get a 50-50 chance of messing up code I didn't write, and with LLMs is even harder to reason the change. It's important to understand scope vs abstraction vs contract when dealing with these changes. For example, I cannot mess with a library I delegated execution to because I am not modifying it, so my current abstraction layer is the problem. With LLMs and all this code I can delegate to my scope is bigger now.

7

1 Comment

I might be late to the blogging party… but I’ve finally started. First post: Getting started - why I’m choosing to write publicly as a dyslexic software engineer and consultant. I’m not even sure who this blog is for yet. Maybe it’s for engineers who think a bit differently. Maybe it’s for those earlier in their careers. Maybe it’s just for future me, looking back and remembering how I moved from one stage to the next. Either way, I’m showing up and pressing publish. Have a read: https://lnkd.in/ex7j3D6J #SoftwareEngineering #Consulting #Dyslexia #LearningInPublic #TechLeadership

22