Mansi 👩‍💻 @MansiCodez

@Alex1here The regret or the job? If it’s the job, was it the hours, the pay, or owning outages you didn’t cause?

Do you regret becoming a software developer? Not the “I hate Jira” kind. The real kind. Weekends gone to prod. Salary looks good until you convert it to hours. AI writes the easy part, you still own the outage. Family thinks you just type. If you could restart at 18, same brain, same market — would you pick this again? Yes / no, and the actual reason.

@antonships Sprint is up there. Sounds fast. Feels like the same ticket with a new number. In standups, “almost done” is the one that hurts.

Words that mean something else if you’re a developer: Cloud Normal people: weather Us: someone else’s computer, and you’re still on-call when it rains Python Normal people: snake Us: the thing that runs until it doesn’t, then it’s a 2-space vs 4-space fight Jira Normal people: a name Us: a graveyard where tickets go to age Cookies Normal people: dessert Us: the reason the site knows you Bug Normal people: insect Us: a feature with worse PR Agile Normal people: flexible Us: standups about why the standup exists Deploy Normal people: send out Us: “please don’t Friday this” What’s the worst one in your team’s vocabulary?

@emilper Fair. Standup isn’t the problem when the people in it can actually unblock each other. It becomes theater when you’re reporting work nobody in the room owns.

My non-tech friend: you guys are so cool, you do standup every day. He meant comedy. I meant 9:30am, camera on, “yesterday I worked on the ticket.” No jokes. No crowd. Just 8 people staring at Jira like it might confess. If software standup was comedy, the only punchline would be “ETA end of day.” What’s the worst standup line you’ve had to say with a straight face?

@YounanixAI Needed. Someone has to review what the agent swore was done.

Vibe coding is wild. No plan. No tests. No idea what the function does. Just: “make it work” “now make it look expensive” “why is prod on fire” And the model says: done. Ship it. Tweet it. Call yourself technical. Then a real bug shows up and the only stack trace is: “it was working in the chat.” Vibe coders, be honest — do you read the diff, or do you just pray the demo doesn’t click the wrong button?

@nickkirt True for personal projects. The second it has users, “it worked in the chat” stops being a strategy.

@reymond______ No mistakes until the first real user.

@shubh19 Exactly. If you can’t explain the diff, you can’t debug it either. The chat won’t be on the call when prod breaks.

@Ra42773Rajesh The interview became a joke when the job changed and the questions didn’t.

Interviewer: reverse a linked list. Me: in production or in this Google Doc? Interviewer: just the logic. Me: okay. I’d ask ChatGPT, then review the edge cases. Interviewer: …without AI. Me: then I’d use a stack. Interviewer: why a stack? Me: because I don’t remember the three-pointer version unless I wrote it this week. Interviewer: so you can’t do it? Me: I can ship it. I can’t cosplay 2015. Interviewer: we’ll get back to you. What’s the most useless interview question you’ve still had to pretend to love?

GraphQL in 30 seconds REST gives you fixed endpoints — you often over-fetch or under-fetch data. GraphQL flips that: one endpoint, and you ask for exactly the fields you need, nested however you want, in a single request. Client asks → server resolves → you get back exactly that shape. No more, no less. Example 👇 Say you want a user's name and their 3 latest post titles. REST (usually 2+ calls): GET /users/42 → { id, name, email, address, ... } GET /users/42/posts → [{ id, title, body, comments, ... }] You get way more fields than you need, across multiple round trips. GraphQL (1 call, exact shape): query { user (id: 42) { name posts(limit: 3) { title } } } Response: { "user": { "name": "Ada", "posts": [ { "title": "Intro to GraphQL" }, { "title": "Why schemas matter" }, { "title": "Resolvers explained" } ] } } One request. Only the fields you asked for. That's the whole pitch.

Write-ahead log, in one story. You run a shop. Customer pays. You could update the ledger later and hope you remember. Or you scribble it in a notebook first: “Paid 500. Order 92.” Then you update the real books. Shop burns down that night. Ledger is gone. Notebook survives. That’s a write-ahead log. Database does the same thing. Write the change to the log Then change the actual data If the server dies mid-write, replay the log No log first = “I think we saved it.” Log first = “we can rebuild it.” That’s why prod doesn’t run on vibes. What’s the database concept that took you way too long to understand?

Recruiter: “We’re a family.” Also the offer: 18 LPA, 6 days, on-call, and “we move fast.” You ask about AI tools. They say, “Yes, we use AI. That’s why we don’t need seniors.” You ask about the interview. DSA. 3 rounds. Design a rate limiter on a whiteboard. You join. First ticket: change a button color. Second ticket: fix the AI PR that passed tests and broke checkout. Month 3 they freeze hiring. Month 4 they ask you to “upskill with AI.” Month 5 they post your role again, cheaper. The job didn’t get replaced. The budget did. Would you take the offer for the brand, or walk for the salary?

Well Architected — quick pillar guide: Operational Excellence → automate everything, learn from failures Security → least privilege, encrypt everything Reliability → design for failure, auto recovery Performance → right tool for the job, monitor everything Cost → pay for what you use, right sizing Sustainability → minimize environmental impact know these cold for the exam 🎯

Offer: 12 LPA Expectations: on-call, system design, AI tools, weekends if prod breaks. Counter: 18 LPA Reply: “Budget is fixed, but you’ll learn a lot.” Same company. Different role. New grad they hired last month: 16 LPA. Less ownership. No on-call. I didn’t lose the negotiation. I just paid extra for experience they don’t want to price. What’s the biggest gap you’ve seen between salary and actual job? Drop the numbers.

Day 27 of #30DaysOfAWS AWS Well Architected Framework — 6 pillars: → Operational Excellence → Security → Reliability → Performance Efficiency → Cost Optimization → Sustainability exam tests these pillars constantly whenever you see "best practice" in a question think: which pillar does this serve? 🎯

Recruiter: 18 LPA, 4 years exp, must know system design, DSA, Kubernetes, and AI agents. Candidate: I have 4 years exp. Current salary 16. Recruiter: Sorry, budget is 12. Freshers with ChatGPT are cheaper. Candidate: Then why the system design round? Recruiter: We still need seniors. Candidate: For 12? Recruiter: We can stretch to 13 if you can also manage the intern. This isn’t a talent shortage. It’s a budget shortage with a senior job description. What’s the most unserious JD you’ve seen this month?