Valerie Pham

This week in vibe coding has been… a ride. ------------------------------------------------------------ Over the past 3 days, I’ve had to revert and re-implement the same codebase 4 times. By now, I know exactly which features I want, what prompts work, and which changes are needed to get there. But here’s the interesting part —��on one particular day it nails it in one shot — like a genius. The next day, it takes me deep into a rabbit hole where nothing works, everything gets over-complicated — a bad developer on a bad day. You get frustrated for sure. And then… it swings back again. It’s almost like it wakes up in a mood. Why does this happen? From a theory perspective — LLMs are inherently stochastic (randomized) unless you lock them down. Slight variations in context or wording can send them down completely different solution paths. The context (baggage it carries) matters. So, how do you keep vibe coding productive (and your sanity intact)? Here’s what’s working for me: 1) Freeze your baseline code after every working iteration — commit like your life depends on it. 2) Lock your prompts — reuse the same structure for recurring tasks. 3) Set temperature low (0–0.2) for more predictable results. 4) Break changes into micro-steps so drift doesn’t snowball. 5) Reset the session if things start feeling “off” — fresh context = fresh start. AND more importantly, When it goes into a rabbit hole → stop there, revert, and give ultra-explicit, “baby steps” instructions: “Go to X file → check this function → change this one line.”. Really you have to act like a caring parent guiding their baby….!! Vibe coding can feel like pair-programming with a genius who sometimes forgets their coffee… but don't worry, if you know how to work with the mood swings, the results can be magical. 

51

11 Comments

I’ve been a SWE for the last 6+ years of my life. If I had to start all over again, I wouldn’t really invest in a course or resources without exploring by myself first. Here are 33 free roadmaps you can use today to start in any field in software engineering & crush it! 1. Backend Roadmap: https://roadmap.sh/backend & Frontend Roadmap: https://lnkd.in/dfk4-pa9 2. Full Stack Roadmap: https://lnkd.in/dRjCeDde 3. JavaScript Roadmap: https://lnkd.in/dA4ucqzz 4. TypeScript Roadmap: https://lnkd.in/dYdd4te4 5. React Roadmap: https://roadmap.sh/react 6. React Native Roadmap: https://lnkd.in/dBG72tqu 7. Vue Roadmap: https://roadmap.sh/vue 8. Angular Roadmap: https://roadmap.sh/angular 9. Node.js Roadmap: https://roadmap.sh/nodejs 10. CS Fundamentals Roadmap: https://lnkd.in/d7iDtkm6 11. GraphQL Roadmap: https://roadmap.sh/graphql 12. Rust Roadmap: https://roadmap.sh/rust 13. Go Roadmap: https://roadmap.sh/golang 14. C++ Roadmap: https://roadmap.sh/cpp 15. Java Roadmap: https://roadmap.sh/java 16. Spring Boot Roadmap: https://lnkd.in/dYRPibvp 17. ASP.NET Core Roadmap: https://lnkd.in/dfFXSAgZ 18. Software Architect Roadmap: https://lnkd.in/dDtNN-ch 19. Software Design and Architecture Roadmap: https://lnkd.in/dUHYkH95 20. System Design Roadmap: https://lnkd.in/d3S79YVJ 21. PostgreSQL Roadmap: https://lnkd.in/d8aTRkBD 22. SQL Roadmap: https://roadmap.sh/sql 23. MongoDB Roadmap: https://roadmap.sh/mongodb 24. AI and Data Scientist Roadmap: https://lnkd.in/dv7Tj4ft 25. Python Roadmap: https://roadmap.sh/python 26. Prompt Engineering Roadmap: https://lnkd.in/diZ2U4dG 27. DevOps Roadmap: https://roadmap.sh/devops 28. Docker Roadmap: https://roadmap.sh/docker 29. Kubernetes Roadmap: https://lnkd.in/dVGxcMKQ 30. Android Roadmap: https://roadmap.sh/android 31. Flutter Roadmap: https://roadmap.sh/flutter 32. UX Design Roadmap: https://lnkd.in/dWXzZgqu 33. Design System Roadmap: https://lnkd.in/dasuSJfZ

1,047

45 Comments

I’ve been a SWE for the last 7+ years of my life. If I had to start all over again, I wouldn’t really invest in a course or resources without exploring by myself first. Here are 33 free roadmaps you can use today to start in any field in software engineering & crush it! 1. Backend Roadmap: https://roadmap.sh/backend & Frontend Roadmap: https://lnkd.in/dfk4-pa9 2. Full Stack Roadmap: https://lnkd.in/dRjCeDde 3. JavaScript Roadmap: https://lnkd.in/dA4ucqzz 4. TypeScript Roadmap: https://lnkd.in/dYdd4te4 5. React Roadmap: https://roadmap.sh/react 6. React Native Roadmap: https://lnkd.in/dBG72tqu 7. Vue Roadmap: https://roadmap.sh/vue 8. Angular Roadmap: https://roadmap.sh/angular 9. Node.js Roadmap: https://roadmap.sh/nodejs 10. CS Fundamentals Roadmap: https://lnkd.in/d7iDtkm6 11. GraphQL Roadmap: https://roadmap.sh/graphql 12. Rust Roadmap: https://roadmap.sh/rust 13. Go Roadmap: https://roadmap.sh/golang 14. C++ Roadmap: https://roadmap.sh/cpp 15. Java Roadmap: https://roadmap.sh/java 16. Spring Boot Roadmap: https://lnkd.in/dYRPibvp 17. ASP.NET Core Roadmap: https://lnkd.in/dfFXSAgZ 18. Software Architect Roadmap: https://lnkd.in/dDtNN-ch 19. Software Design and Architecture Roadmap: https://lnkd.in/dUHYkH95 20. System Design Roadmap: https://lnkd.in/d3S79YVJ 21. PostgreSQL Roadmap: https://lnkd.in/d8aTRkBD 22. SQL Roadmap: https://roadmap.sh/sql 23. MongoDB Roadmap: https://roadmap.sh/mongodb 24. AI and Data Scientist Roadmap: https://lnkd.in/dv7Tj4ft 25. Python Roadmap: https://roadmap.sh/python 26. Prompt Engineering Roadmap: https://lnkd.in/diZ2U4dG 27. DevOps Roadmap: https://roadmap.sh/devops 28. Docker Roadmap: https://roadmap.sh/docker 29. Kubernetes Roadmap: https://lnkd.in/dVGxcMKQ 30. Android Roadmap: https://roadmap.sh/android 31. Flutter Roadmap: https://roadmap.sh/flutter 32. UX Design Roadmap: https://lnkd.in/dWXzZgqu 33. Design System Roadmap: https://lnkd.in/dasuSJfZ

857

86 Comments

My next piece of advice for myself and colleagues affected by layoffs - Affix your own oxygen mask first. Preflight safety information on commercial airlines emphasizes putting on your own oxygen mask before helping others. The same is true of other stressful events. You can't support others if you aren't also taking care of yourself. You work here, so chances are your familie's economic security is your responsibility. Taking care of yourself first isn't selfish so make sure you're getting enough sleep, exercise, and time with loved ones. Regulate your substance consumption. With that foundation set you can go out and be absolutely unrelenting in your search for your next role. That doesn't mean we turn looking for our next opportunity into the Hunger Games. Many of you are pursuing the same job openings as the person sitting next to you but it's not a zero-sum game. The pursuit makes us competitors, not adversaries. You should be applying to any roles that you're excited about, even if you know that others are, too. Focus on opportunities that align with your skills and goals and be generous by sharing the ones that don't match up. Above all, don't feel guilty about landing a job before others do. And don't burn bridges by talking negatively about former teams or teammates. Hiring managers and interviewers treat negativity like nuclear waste and badmouthing your former manager, colleagues or team isn't going to give you a leg up. Focus on what makes you awesome and the strengths you bring to the table. Pursue opportunities with the confidence and professionalism I know you are all capable of. Nobody's success diminishes your own. I look forward to seeing each and every one of you landing where you want to be. Take care of yourself first and everything else will follow.

45

1 Comment

I’ve been building some greenfield software lately, and it’s exhilarating. Most engineers rarely get to build end-to-end. At work, we often inherit, maintain, and optimize things we didn’t create. But when you take something from idea to implementation, you learn fast. The ownership, the clarity, and the momentum, next level stuff! So my pitch is: Just build something. Not a tutorial. Not a clone. Something real. Something that matters. Something yours. It’ll change how you think.

138

34 Comments

Hiring engineers for hard operational problems is different. Most candidates can write clean code. Fewer can sit with a dispatcher at 7am and understand why their shift planning breaks down. Fewer still can go back to the codebase and design something that actually works in production with those constraints. What I look for at Mirage Metrics is not just skill with Python, SQL, or distributed systems. It is a willingness to go into environments that are messy, sometimes chaotic, and keep digging until the real bottleneck shows up. That can mean inventory mismatches hidden in Excel, planning rules written on a whiteboard, or API endpoints that return inconsistent data depending on who queries them. The best engineers I have worked with are the ones who do not flinch at this. They treat debugging a customs declaration pipeline with the same seriousness as debugging a memory leak. They know that solving “boring” problems in data quality or workflow mapping is what unlocks everything else. On the technical side, I test for clarity. Can someone explain the trade-offs between running a model via an API versus hosting it on GPUs without getting lost in jargon. Can they design a schema validator that will still make sense six months later when requirements change. Can they write glue code that is robust, not fragile. On the personal side, I look for stamina. These projects are rarely about building a shiny feature in isolation. They require iteration with operators, late-night debugging of OCR failures, and the patience to integrate into legacy systems that were never designed for AI. If you want to work on AI for logistics, manufacturing, or mining, the question is not only whether you can code. It is whether you can hold your ground in the real world, in front of the people whose work depends on your system. That is the bar.

17

5 Comments

As more engineers use coding agents professionally, the traditional junior engineering mentorship model delays two skills they need to be successful with agents. Junior engineers need earlier introductions to task definition and code review. I wrote about that here: https://lnkd.in/gKgv67NC I'm reticent to write about something so new to our field, but the gap is clear and these tools are being adopted quickly. Given the loud opinions in the space, people who write software professionally have a lot to think about, and introspecting our mentorship process seems like a good start!

32

2 Comments

The industry works on a funnel of experience that takes junior/mid level engineers as input and produces senior engineers as an output. With all the buzz around AI replacing junior/mid level engineers, not sure how the funnel will continue to work ¯\_(ツ)_/¯

22

2 Comments

In 9 months, the best tool in my toolkit as a software engineer went from singleline autocomplete to multiline autocomplete to CLI-based coding agents to Slack-based coding agents. The creative destruction in this space is one hell of a rollercoaster! Recently, I have gone full throttle on agentic coding. That is, putting an LLM in the driver’s seat when writing code while I navigate in the passenger’s seat. And I think there is no going back. Here are 9 miscellaneous observations over the past 9 months: 1) Expect your productivity to decrease at first. The best way to learn how to work with LLMs is to work with LLMs but they are really not good at some things. You will only know to avoid the agentic traps like it infinitely fixing its own errors by first falling into the traps. 2) Don’t buy an annual subscription. I’ve moved more time from Windsurf to Claude Code but I’m still stuck with 6 months left of a 12 month Cursor subscription I don’t touch anymore. 3) Pick a tool that is token unconstrained. I can really tell the difference between tools that charge by month and tools that charge by token. If your employeer values your time, the cost doesn't matter. 4) LLMs love context, so give it to them. An early mistake I made was writing one sentence and expecting my bugs to be squashed. Now, I write paragraphs and pages. 5) The fastest way to give context is to ramble using a dictation tool. But if you work in an office, you will look crazy. 6) Improvements to your docs are now multiplied. LLMs never learn and always make the same mistakes unless you write your lessons down somewhere. 7) Improvements to your developer experience are now multiplied. Previously, I would manually run a linter to stop LLMs spending too long fixing linting errors line-by-by. Now, I get a PR that passes CI. 8) An under-discussed skill is knowing when to go back to artisanally writing code by hand. 9) Everyone who raves about agentic coding always writes their blog posts by hand. I hope the last 2 points give hope to anyone who fears LLMs will take everyone’s jobs.

89

16 Comments