It’s an eMacs package that starts from a spec and will help you iterate on it if you want, and works from a core set of docs about a project plus active specs.
It also will keep a treesitter based representation of your codebase to help you form the architecture. It has no write tools, it will read a diff of your code on save to help. It also can be prompted openly.
It’s a pair programmer, but the other way around versus traditional agent harnesses. You’re the coder, the AI watches
Its knowledge mainly, my company is covering tuition and I’ve always had a desire to understand computing “all the way down.” If the outcome would be that I’m still working similar jobs, I’d be alright with that. I commented on a different parent about what coursework I’m considering, maybe that would help me break into something lower level and not just be another backend job?
Super valuable advice. The program I’m considering has VLSI design, chip architecture, and FPGA work all at the graduate level. I don’t know if the school has fabs but it’s seemingly well known and well funded so maybe?
As a part of this degree I’ve also selected coursework that is more related to CS. I’m looking at a compilers class, OS class, and neural networks as well.
With a focus on FPGA do you think that’s broadly applicable and evenly distributed geographically? I am prepared to move for certain industries, like working in space technology
This is exactly where I’m at with my CS career. I am tired of abstraction. I want to be able to strip it away and work at the lowest levels of the stack. I’ve worked in embedded real-time environments before and found working with the system constraints to be highly rewarding. Yes there are tools that can write and test code for us now, but it’s also true that someone still has to build the low level high performance stuff.
What opportunity do you see in the realm of atoms?
This was an amazing read, and frankly inspiring with regard to your final thoughts. I’ve yet to build anything substantial on the side but still searching for that idea
now I’m convinced that a certain senior engineer at a very well known tech company had this as an instruction manual. I’ve experienced each of these multiple times from the same person!
On my own though, I’m certainly guilty of the 100 round trips. I tend to mentally group nits and not spell out every one that I find
You: Have problems related to data, want to modernize your operations (no more clickops), want to build a mobile app, want to add AI to your product.
Me: 10 years of experience, previously at Amazon, currently an LLM research engineer. Expert in CI/CD and Infrastructure as Code through Terraform. Skilled in building AI-enabled application through the usage of RAG and agent-based workflows. Also have built many MVP style products with Flutter and .NET.
I can handle your Sagemaker MLOps, can help experiment with LLMs, and can build products with Flutter, or Ruby on Rails. Fullstack. I've helped one startup reach exit by building fitness tracker integration into their flutter app (https://fytfeed.com). I've tried to build several products myself with C#/.NET, Flutter, and Django
It may be time to finally learn it. I've standardized on Python, Clojure and C# of all languages (Julia for fun) but Rust may help me dive deeper into systems level problems.
You're right. I think my biggest problem is that I always assume these crazy OSS projects were written because the author has always been this programming genius, when in reality they were just following their interests. For example, I caught myself with the proxy thinking about production scale. Who cares about production?? I just need to get going and build a single-threaded proxy that gets the job done. No need to worry about complexity, it's just for me.
I loved this article, this is exactly what programmers in my community complain most about with regard to lisp.
The greatest strengths for me are the scoping and how lisp enables interactive development. For one, I don't even see parens anymore when writing code. They've become automatic where I'm able to parse structure at least better than I have in the past. With a REPL, (I specifically use Clojure), development has become fun again. A typical debugging process is attaching to my actual program, and calling functions seeing what happens.
I love the flow of defining functions in my code, `def`-ing some test data, and calling those functions. I'll use `cider-inspect` to see if the data structure looks correct and iterate.
Everyone should at least try a lisp. Iterative, interactive development is so fun and powerful.
I'm wrestling with this same question myself. I have never had any idea. For me I've started businesses when the external pressure of having a co-founder dictated that I should. Or, I'd be in a fit of passion-inspired development for a few weeks until it was time to talk to customers. I think the answer varies from person to person. It's hard to say what a good heuristic would be. We all operate with different heuristics.
For instance I now have a family, and starting a business would either take me away from family time, involve nights and weekends work, or involve quitting my day job (of which the bar is now very high).
Valuable advice. I've found the same w.r.t. working at a small company versus an enterprise. I think I'm in the niche carving stage of my career but I have no idea how to sell myself. I'm in the same boat, I can and do operate across that wide swathe of experience as well.
It's been feeling lately like I'm at a crossroads, I either settle down and "specialize" by picking a stack and an "area," or I build my own product.
I've had a lot of trouble with hiring lately due to being a generalist. I've spent 10 years as a developer but not more than 2-3 at a time in a specific "stack." This has hurt me more times than I can remember.
However, I've built amazing things for amazing companies when I do get jobs. I've architected and built end-to-end products by myself and within teams. A few projects I've worked on: a web scraping product, data labeling software for machine learning, commercial eyetracking software, a custom learning management system, machine learning models from scratch, high volume image processing, etc. In each of these projects too I did end-to-end work. Development, deployment, infrastructure, ops.
Each one of these products though was with a different stack. JS, C#, Clojure, C++, Java, Python, the list goes on.
Common feedback I've received is that I'm not experienced enough to deliver in a role. Anecdotally, engineers that I talk to where I'm located have spent their entire careers using only JS, building Next.js sites or whatever the current hotness is. Not that there's anything wrong with that, but it does seem like the landscape has shifted to the specialists.
Happy 2024 :). I've gotten so much out of this site. 2024 marks 10 years for me as an SWE officially. Starting with a soulless internship in a defense contractor, working at many early stage startups, and ending with a FAANG.
The one constant? This site. I never post, but I hope to more. Love the productive discussions that I always see on this site. It's a breath of fresh air after spending any time on X, LinkedIn, Facebook, etc.
https://github.com/jaketothepast/codetutor
It’s an eMacs package that starts from a spec and will help you iterate on it if you want, and works from a core set of docs about a project plus active specs.
It also will keep a treesitter based representation of your codebase to help you form the architecture. It has no write tools, it will read a diff of your code on save to help. It also can be prompted openly.
It’s a pair programmer, but the other way around versus traditional agent harnesses. You’re the coder, the AI watches