HN user
https://osequi.com
766fc7
Exactly! What I use is a main workflow document where I embed at every step pointers to architecture and templates.
My prompt is ... "We are implementing the X feature. We are at step 6. Plan first"
Then the agent spits out identical plans then identical code for every feature.
In my 30+ years of SWE/SWA career this is the first time I can harvest the benefits of a well defined and exactly implemented architecture.
Thanks to LLMs.
Before LLMs even if the architecture principles were simple and clear, distilled into templates + codegens added for boilerplate / skeleton generation ... It was impossible to follow them on the long run. Devs tried their best, but on the long run everything eroded and there were no resources for refactoring.
Now, with coding agents, I was able to create a production grade app following a similar architecture to Presentation Domain Data Layering, from this article.
Now the codebase is 100% uniform both in content (code) and structure (files and folders). It's like being written by a single person. Finding a specific file takes a second with no cognitive load. Editing a file is straightforward since every file follows a specific template.
LLMs have benefits and drawbacks, and in this case their help is enormous.
Good point! Reviewing code, in the AI era, in my practice, means skimming code and looking for patterns.
I use templates / conventions and make the AI generate code using them. When reviewing code I'm scanning if a file uses a specific template and follows specific conventions.
This can't catch subtle errors like a function is re-created vs is re-used (duplicate code), unnecessary code (bloat), inconsistent naming (a Button component has a cssRow() styling function associated vs cssButton()).
When you start editing, using code these little things add up, consume your attention, drain you up, giving no flow and resulting in minimal productivity.
This specify-encode-fulfill loop/method is effective to make agents create bug-free code.
In my version of this workflow I do specify myself, then let the LLM do the rest.
This way 1.) I'm 100% sure the understanding/spec is good 2.) It's translated into an executable format so the implementation can be verified 3.) The implementation has maximum code coverage tests which steers the AI to produce code which follows standards, fits into the existing codebase, and it's very easy to refactor.
So far, this is the one and only advantage of using LLMs in my SWE practice. They glue together (human written) specs with code, with confidence, in no time.
The question is whether AI / LLMs gets better.
I'm not an ML expert, but regarding code _quality_ I see no progress at all in the last couple of years. LLMs still write code by using probabilistic calculations vs. applying rigorous thinking and logic.
This is only good while no one has to look under the hood. When trying to understand and fix code written by LLMs you'll realize what a mess they produce. It's a codebase without any systematic thinking inside. Everything is ad-hoc, wired together to pass the tests, and to conform to some templates. No deliberate practice, no intelligence at all in the code.
This can't be a long term strategy for an entire industry.
It's a gut feeling.
We _know_ LLMs can't be _that_ good as they are promoted.
I've spent the last 6 months creating a production grade app from scratch with Claude where I wrote no single line of code. I've reviewed code and it was looking good, almost completely following my templates, workflows, skills.
Now I've started to make minor manual updates and I'm horrified. Claude has no idea why there were those templates and instructions in place. It followed them blindly without grasping their spirit. The end result is like a very junior dev copy-pasting answers from Stack Overflow into the codebase. No consistency, chaotic application of different conventions, duplicated code, ghost code (does nothing), and perhaps more as I'm digging in.
The pros: The code works, all tests pass (43% code / 57% tests, 1:1.3 ratio), the UI looks good with visible glitches
The cons: I'll have to rewrite most of the code on the long run, make it fit, easy to maintain.
The verdict: I wouldn't started this project alone. Claude get me through to v0.1.0 / MVP where I've focused solely on the product: technologies, architecture, functionality, and usability. Now it's easier to refactor all for v0.2.0 manually without Claude.
So this might be our gut feeling: we know it's something good, but not as good as the stakeholders might promote. We know it helps in some ways but it's a nightmare in other ways.
We are not anti-AI but rather pragmatic: Not that AI enthusiasts we are expected to be.
SEEKING WORK | Europe | Remote
I wear multiple hats: Researcher / Senior Software Architect / React Lead / Design Engineer / Junior AI Engineer.
I can help you with your AI software engineering strategy and implementation with a focus on correctness: https://www.osequi.com/studies/list/list.html
- Résumé/CV: https://osequi.com/
- Email: bartus.csongor@gmail.com
Senior Software Architect | React Lead | Design Engineer | Remote, EU
- Location: Europe
- Remote: Yes
- Willing to relocate: Maybe
- Résumé/CV: https://osequi.com/, https://chat.osequi.com/ (AMA with AI)
- Email: bartus.csongor@gmail.com
I deliver better software, faster:
- I solve the two top pain points of JS / TS / React [1]
- Using lightweight formal and semi-formal methods [2]
- I integrate multiple disciplines (entrepreneurship, math, computer science, UX/UI design) to create better products [3]
I'm interested in companies with a high societal impact: synthetic biology, education, healthcare, personal growth, financial stability.
[1] https://2023.stateofjs.com/en-US/usage/#top_js_pain_points
SEEKING WORK | Europe | Remote
I wear multiple hats: Researcher / Senior Software Architect / React Lead / Design Engineer / Junior AI Engineer.
I can help you with your AI software engineering strategy and implementation with a focus on correctness: https://www.osequi.com/studies/list/list.html
- Résumé/CV: https://osequi.com/
- Email: bartus.csongor@gmail.com
Senior Software Architect | React Lead | Design Engineer | Remote, EU
- Location: Europe
- Remote: Yes
- Willing to relocate: Maybe
- Résumé/CV: https://osequi.com/, https://chat.osequi.com/ (AMA with AI)
- Email: bartus.csongor@gmail.com
I deliver better software, faster:
- I've created a novel methodology to produce likely-correct software using formal and semi-formal methods [1]
- I integrate multiple disciplines (entrepreneurship, math, computer science, UX/UI design) to create better products [2]
- I use LLMs to generate software based on mathematically correct diagrams [3]
I'm interested in companies with a societal impact: improving lives through education, healthcare, personal growth, financial stability.
[1] https://www.osequi.com/studies/list/list.html
I'm writing a study about how to write likely-correct studies ... :)
This is a second part of a series on likely-correctness, the first is how to create likely-correct software: https://www.osequi.com/studies/list/list.html
Here is a birds-eye view of programming (classic, functional, quantum) vs category theory vs logic -- aka the computational trilogy:
https://ncatlab.org/nlab/show/computational+trilogy
It helped me a lot putting into context my existing programming knowledge while learning category theory
Just learnt Applied Category Theory, it's a very big fun, hope you're enjoying it too :)
That's a backend issue I guess ...
What happened to the very elegant GraphQL? Where the client _declares_ its data needs, and _that's all_, all the rest is taken care by the framework?
Compared to GraphQL, Server Components are a big step back: you have to do manually on the server what was given by default by GraphQL
SEEKING WORK | Europe | Remote
I wear multiple hats: Researcher / Senior Software Architect and Engineer / Design Engineer and recently Junior AI Engineer.
I can help you with your AI software engineering strategy and implementation with a focus on correctness: https://www.osequi.com/studies/list/list.html
- Résumé/CV: https://osequi.com/
- Email: bartus.csongor@gmail.com
Senior Software Architect | Junior AI Engineer | Remote, EU
- Location: Europe
- Remote: Yes
- Willing to relocate: Maybe
- Résumé/CV: https://osequi.com/
- Email: bartus.csongor@gmail.com
Mathematics, Computer Science and UI/UX design specialist with 25+ years of experience.
I solve two of the most challenging software engineering problems: code organization and state management, using lightweight formal methods -- Applied category theory, Finite state machines -- to bring academic rigor to the full stack: https://www.osequi.com/studies/list/list.html
I'm a junior AI Engineer versed in prompting, agentic development with Vercel AI SDK. Finishing my first AI app, an agent/assistant helping to write likely-correct studies, articles and technical documents.
Seeking:
- Teams building better software, faster
- Particularly interested in fintech, health-tech, and developer tools
Formalization, correctness is about models. [1]
There are formal methods where the underlying model is mathematically sound. There are semi-formal methods where the underlying model is structured but not proven to be sound.
For example, in your case ("organizes knowledge") a formal method is ologs from category theory. That method assures that the concepts and their relationship in your knowledge base are mathematically correct.
When you want to transfer that knowledge base into a documentation system you might want look for a mathematically sound model, but I'm afraid there is no such model, so what's left is a semi-formal method / a likely-correct model.
Right now I'm building such a likely-correct model for writing, contact me for more info.
[1] https://www.osequi.com/slides/formalism-correctness-cost/for...
Perhaps there is no such thing like absolute truth.
In category theory / ologs, a formal method for knowledge representation, the result is always mathematically sound, yet ologs are prefixed with "According to the author's world view ..."
On the other way truth, even if it's not absolute, it's very expensive.
Lately AWS advocates a middle ground, the lightweight formal methods, which are much cheaper than formal methods yet deliver good enough correctness for their business case.
In the same way MIT CSAIL's Daniel Jackson advocates a semi-formal approach to design likely-correct apps (Concept design)
It seems there is a push for better correctness in software, without the aim of perfectionism.
I've recently created a likely-correct piece of software based on these principles.
https://www.osequi.com/studies/list/list.html
The structure (ontology, taxonomy) is created with ologs, a formal method from category theory. The behavior (choreography) is created with a semi-formal implementation (XState) of a formal method (Finite State Machines)
The user-facing aspect of the software is designed with Concept Design, a semi-formal method from MIT CSAIL.
Creating software with these methods is refreshing and fun. Maybe one day we can reach Tonsky's "Diagrams are code" vision.
I have a good understanding of how to create likely-correct software: https://www.osequi.com/studies/list/list.html
Now I'm learning AI/LLMs from the perspective of correctness. So far I have two 'maxims' to guide me:
- AI shines where humans struggle (for prompt engineering)
- An LLM is nothing but an API call (for software engineering with AI)
SEEKING WORK | Europe | Remote
A Senior / Lead React Front-end Developer, Design Engineer, Software Architect is looking for a job.
I have dual skills: a degree in Computer Science and UI/UX design works featured in online galleries.
In the last 15 years I've been designing and developing websites and applications. From back-end to front-end, from architecture design to UI/UX design, on different stacks. I've built and managed front-end teams, shaped products.
Currently, I specialize in React and TypeScript, offering expertise in:
- Team Leadership: Building and guiding React teams, producing high-quality code even with junior developers.
- Design Engineering: Developing robust design systems, enabling rapid prototyping, shaping product design, and bridging developer-designer communication gaps.
- React Software Architecture: Solving pain points like state management and code architecture: https://www.osequi.com/studies/list/list.html
- AI-Driven Development: Advising on building likely-correct React apps and training AI models for React app generation.
Résumé/CV: http://metamn.io/
Email: bartus.csongor@gmail.com
Senior Software Architect | Formal Methods, UI/UX Design | 25+ Years | Remote
- Location: Europe
- Remote: Yes
- Willing to relocate: Maybe
- Résumé/CV: https://osequi.com/
- Email: bartus.csongor@gmail.com
Mathematics, Computer Science and UI/UX design specialist with 25+ years of experience.
I solve two of the most challenging software engineering problems: code architecture and state management, using formal methods (Applied category theory, Finite state machines) to bring academic rigor to the full stack: https://www.osequi.com/studies/list/list.html
I also facilitate product design by bridging design and development, leveraging my deep understanding of both visual design and its technical implementation.
Achievements:
- Built and led engineering teams (30+ engineers)
- Co-founded startups (1 successful exit)
- International consulting (Silicon Valley, UK, EU, Singapore)
- Featured in UI/UX design galleries
- Recently completed two-year R&D sabbatical focused on software correctness and rapid iteration methodologies
Seeking:
- Senior / Principal architect or technical leadership roles at companies valuing correctness, maintainability, and design excellence
- Joining teams building better software, faster
- Particularly interested in fintech, health-tech, and developer tools
Working on creating likely-correct software with formal and semi-formal methods for rapid iteration.
Done the first demo: https://www.osequi.com/studies/list/list.html, now focusing on "diagrams as code": https://tonsky.me/blog/diagrams/
SEEKING WORK | Europe | Remote
A Senior / Lead React Front-end Developer, Design Engineer, Software Architect is looking for a part time, fractional job.
I have dual skills: a degree in Computer Science and UI/UX design works featured in online galleries.
In the last 15 years I've been designing and developing websites and applications. From back-end to front-end, from architecture design to UI/UX design, on different stacks. I've built and managed front-end teams, shaped products.
Currently, I specialize in React and TypeScript, offering expertise in:
- Team Leadership: Building and guiding React teams, producing high-quality code even with junior developers.
- Design Engineering: Developing robust design systems, enabling rapid prototyping, shaping product design, and bridging developer-designer communication gaps.
- React Software Architecture: Solving pain points like state management and code architecture, as highlighted in the 2023 State of JS survey: https://2023.stateofjs.com/en-US/usage/#top_js_pain_points.
- AI-Driven Development: Advising on building likely-correct React apps and training AI models for React app generation.
I prefer a part-time or fractional role as I also run a research and development studio focused on correct software: https://www.osequi.com/.
Senior / Lead React Front-end Developer, Design Engineer, Software Architect -- Part time, fractional job
- Location: Europe
- Remote: Yes
- Willing to relocate: Maybe
- Technologies: React, TypeScript, Next.js, Design systems, XState, Functional software architecture
- Résumé/CV: http://metamn.io/
- Email: bartus.csongor@gmail.com
I have dual skills: a degree in Computer Science and UI/UX design works featured in online galleries.
In the last 15 years I've been designing and developing websites and applications. From back-end to front-end, from architecture design to UI/UX design, on different stacks. I've built and managed front-end teams, shaped products.
Currently, I specialize in React and TypeScript, offering expertise in:
- Team Leadership: Building and guiding React teams, producing high-quality code even with junior developers.
- Design Engineering: Developing robust design systems, enabling rapid prototyping, shaping product design, and bridging developer-designer communication gaps.
- React Software Architecture: Solving pain points like state management and code architecture, as highlighted in the 2023 State of JS survey: https://2023.stateofjs.com/en-US/usage/#top_js_pain_points.
- AI-Driven Development: Advising on building likely-correct React apps and training AI models for React app generation.
I prefer a part-time or fractional role as I also run a research and development studio focused on correct software: https://www.osequi.com/.
SEEKING WORK | Europe | Remote
A Senior / Lead React Front-end Developer, Design Engineer, Software Architect is looking for a part time, fractional job.
I have dual skills: a degree in Computer Science and UI/UX design works featured in online galleries.
In the last 15 years I've been designing and developing websites and applications. From back-end to front-end, from architecture design to UI/UX design, on different stacks. I've built and managed front-end teams, shaped products.
Right now I focus on React and I can be helpful in a couple of ways:
- Building, guiding, leading React teams to produce high quality code even with junior developers
- Design Engineering: Building robust design systems, rapid prototyping new ideas or features, shaping product design, developer - designer communication
- React Software Architecture: Solving top pain points like code architecture and state management (https://2023.stateofjs.com/en-US/usage/#top_js_pain_points)
I prefer a part-time / fractional job because I run a research and development studio focusing on correct software: https://www.osequi.com/
If you are interested in likely-correct React apps, or planning to train your AI to generate such React apps, then I'm available full time.
bartus.csongor@gmail.com
Senior / Lead React Front-end Developer, Design Engineer, Software Architect -- Part time, fractional job
- Location: Europe
- Remote: Yes
- Willing to relocate: Maybe
- Technologies: React, TypeScript, Next.js, Design systems, XState, Functional software architecture
- Résumé/CV: http://metamn.io/
- Email: bartus.csongor@gmail.com
I have dual skills: a degree in Computer Science and UI/UX design works featured in online galleries.
In the last 15 years I've been designing and developing websites and applications. From back-end to front-end, from architecture design to UI/UX design, on different stacks. I've built and managed front-end teams, shaped products.
Right now I focus on React and I can be helpful in a couple of ways:
- Building, guiding, leading React teams to produce high quality code even with junior developers
- Design Engineering: Building robust design systems, rapid prototyping new ideas or features, shaping product design, developer - designer communication
- React Software Architecture: Solving top pain points like code architecture and state management (https://2023.stateofjs.com/en-US/usage/#top_js_pain_points)
I prefer a part-time / fractional job because I run a research and development studio focusing on correct software: https://www.osequi.com/
If you are interested in likely-correct React apps, or planning to train your AI to generate React apps, then I'm available full time.
This fully resonates with me.
When the AI boom started in 2022 I've been already focused on how to create provably, or likely correct software on budget.
Since then, I've figured out how to create correct software fast, on rapid iteration. (https://www.osequi.com/)
Now I can combine productivity and quality into one single framework / method / toolchain ... at least for a niche (React apps)
Do I use AI? Only for pair programming: suggestions for algorithms, suggestions for very small technical details like Typescript polymorphism.
Do I need more AI? Not really ...
My framework automates most part of the software development process: design (specification and documentation), development, verification. What's left is understanding aka designing the software architecture, and for that I'm using math, not AI, which provides me provably-correct translatable-to-code-models in a deterministic way. None of these will be offered by AI in the foreseeable future
SEEKING WORK: Senior / Lead React Front-end Developer, Design Engineer -- Part time
- Location: Europe
- Remote: Yes
- Willing to relocate: Maybe
- Technologies: React, TypeScript, Next.js, Design systems, XState, Functional software architecture
- Résumé/CV: http://metamn.io/
- Email: bartus.csongor@gmail.com
I have dual skills: a degree in Computer Science and UI/UX design works featured in online galleries.
In the last 15 years I've been designing and developing websites and applications. From back-end to front-end, from architecture design to UI/UX design, on different stacks. I've built and managed front-end teams, shaped products.
Right now I focus on React and I can be helpful in a couple of ways:
- Leading, guiding React teams to produce high quality code
- Designer / Developer communication
- Design Engineering: Building robust design systems, rapid prototyping new ideas or features, shaping product design
I prefer a part-time job because I run a research and development studio focusing on correct software: https://www.osequi.com/
If you are interested in likely-correct React apps, or planning to train your AI to generate React apps, then I'm available full time.