Topic 4 — The language of tech

Goal: Learn the everyday tech vocabulary so you can follow any conversation.

Lesson 4.1 — The locked door that's propped open

Teodora Vrabie is sitting in on her first real tech meeting. Not a job — Bao Trinh, a product manager at Brightwell, let her dial into a planning call as a favor, the kind of thing that happens when you ask people for coffee enough times. Teo taught high-school geography for nine years. She has a notebook open and a growing list of words she doesn't understand: the backend's flaky so we're rate-limiting the API before we ship the MVP, then we'll iterate on the metric. Words she half-recognizes, in one sentence, said at full speed by people who weren't trying to confuse her.

This is the moment a lot of career-changers decide tech isn't for them. The vocabulary sounds like a locked door, and the lock looks like it needs a computer-science degree.

It doesn't. The thing nobody tells you on day one: almost all tech jargon is a simple idea wearing a short name. "Rate-limiting" means slow down the requests so we don't overload it. "Iterate" means improve it in small steps. The words are shorthand, so busy people can say a paragraph in a syllable. Once you know what each one unpacks into, the locked door turns out to have been propped open the whole time.

Jargon is a vocabulary list, not a knowledge test. And the list is short.

And you don't need the deep version. Teo will never write the code that rate-limits an API. Her job — anyone's job who isn't an engineer — is recognition: hear the word, know roughly what it points at, and keep following instead of freezing. A few dozen terms cover most of what gets said in a standup, an interview, or a Slack channel. This topic is that list, with the cast of Brightwell to make it stick.

Lesson 4.2 — The eight words that unlock the room

Start with the technical core. These eight come up constantly, and Teo learned to spot every one of them within a week of paying attention.

The first is the cloud. When Marisol Ferreira, Brightwell's founder, says "our app runs in the cloud," she doesn't mean the sky. She means the app and its data live on powerful computers — servers — that someone else owns and maintains, reached over the internet, instead of on your own laptop. "In the cloud" = "running on remote servers, accessed online."

Brightwell sells its study-planner as SaaS — Software as a Service. You don't buy a box and install it; you pay a subscription and use it over the internet. Gmail is SaaS. Netflix is SaaS. The term means software you rent and reach online rather than own.

Then the one that scares newcomers most: API — Application Programming Interface — simply how two pieces of software talk to each other and pass data back and forth. The standard picture, and a good one, is a waiter in a restaurant. You (one program) don't march into the kitchen; you tell the waiter what you want, the waiter carries the order to the kitchen (another program) and returns with the dish. When Brightwell's app shows a student their school calendar, it asks the school's system through its API and gets the dates back. You ask, it answers, you never see the kitchen.

A few more that travel together:

• Platform — a base others build on top of. An app store is a platform; so is a service lots of products plug into.
• Frontend / backend — the frontend is the part you see and click; the backend is the behind-the-scenes engine and logic. In the restaurant: the dining room is the frontend, the kitchen is the backend.
• Database — organized storage, so data can be saved and pulled back later. Every student record at Brightwell lives in a database.
• Bug vs. feature — a bug is the software not working as intended; a feature is a thing it's supposed to do. "Is that a bug or a feature?" is asking broken, or intended?

Eight words. Learn these and you can follow most of any technical conversation without reading a line of code.

Lesson 4.3 — How the work gets talked about

The second cluster describes how the team builds the product, and Teo hears these even more often than the technical ones, because most of a meeting is people coordinating, not coding.

Several you already met in Topic 3, so here they're just nameplates: teams work in agile style, in short cycles called sprints, often using a method named scrum, pulling from a prioritized to-do list called the backlog, checking in at a quick daily standup. The smallest first version they release to test an idea is the MVP (minimum viable product).

The new ones for this topic:

• Roadmap — the plan of what's coming and roughly when. Not a contract, a direction. When Bao says "that's on the roadmap for Q3," he means we intend to build it around then.
• Ship / launch / release — all three mean put it out to users. "We're shipping Friday" = it goes live Friday. (You'll hear "ship" the most; engineers love it.)
• Iterate — improve through repeated small changes rather than one giant rewrite. Launch a rough version, watch what happens, adjust, repeat.

Davor Halász — a former accountant, now a junior data analyst one step ahead of Teo on the same career change — gave her the trick that made these click. "Don't translate them in your head mid-sentence. You'll fall behind. Let them wash over you for a week. By Friday 'we're iterating on the MVP before the next sprint' will sound like a normal sentence, because it is one." The words stop being foreign roughly the moment you stop bracing against them.

Lesson 4.4 — Talking about whether it worked

The third cluster is the language of the business — money, measurement, and the people who care. This is what makes Teo sound like she understands why a company does what it does, beyond what it builds.

A metric is any number you track. A KPI — key performance indicator — is a metric that actually matters for the goal, the headline number. Brightwell watches active users (how many people use the app), revenue (money in), and churn (the share of customers who cancel). Churn keeps Marisol up at night: a subscription business lives or dies on whether people stay.

ROI — return on investment — asks the oldest question in business: was it worth it? Value gained versus what it cost. Spend three engineer-months on a feature that barely moves churn, and the ROI was poor, however nice the feature.

A stakeholder is anyone with a stake in a decision — Marisol, the engineers, sales, sometimes a big school customer. "Let's check with stakeholders" means the people this affects should weigh in.

Scale / scalable is about growth. To scale is to handle many more users; "scalable" means the system can grow to ten times the load without falling over. When Brightwell raised money, the pitch was that the product could scale — add a thousand more schools without rebuilding everything.

And you'll hear the funding words — startup, scale-up, VC (venture capital) — which describe a company's stage and how it raises money. Topic 5 takes those apart properly; for now, just know they're about where a company is in its life and who's backing it.

A metric is any number you watch. A KPI is the number you'd defend in a meeting.

Lesson 4.5 — The 2026 words: AI, LLMs, and not bluffing

One cluster didn't exist in most beginner guides a few years ago and now comes up in nearly every interview Renske Aldous, Brightwell's recruiter, runs. AI.

AI is artificial intelligence — software doing tasks we used to think needed a human. The kind behind the tools everyone's talking about is an LLM, a large language model: the engine inside ChatGPT (OpenAI), Claude (Anthropic), and Gemini (Google). An LLM is trained on enormous amounts of text and predicts likely words, which is why it can write, summarize, and answer in fluent language.

Two more words you'll really need:

• Prompt — the instruction you give the model. "Write a summary of this report" is a prompt. Better prompts get better results; that's the whole skill people mean by "prompting."
• Hallucination — when the model states something false with total confidence. Because these models are built to always produce a fluent answer, they'll sometimes invent a fact, citation, or number that sounds right and isn't. It's not lying; there's no intent. It's the model guessing and presenting the guess as fact. The practical lesson: useful tool, always verify anything that matters.

You'll also hear about copilots or AI assistants — AI features now baked into everyday tools (your email drafting replies, your code editor suggesting code, your docs summarizing themselves). Renske told Teo she doesn't expect a career-changer to be an AI expert. She expects them to know what an LLM is, to have actually used one, and not to look blank when "hallucination" comes up. That's the bar, and it's a low one to clear.

Which brings up the rule that governs all of this vocabulary: using a term right signals you belong; using one wrong signals the opposite, loudly. Misuse "API" to mean "the website" and an interviewer notices. So if you're not sure, don't bluff. Recognize the word, and ask. "When you say platform here, do you mean the app store, or our internal system?" reads as sharp and engaged, never as ignorant. Curiosity is the most credible thing in the room.

Worked example — Teo decodes one real sentence

Six weeks in, Teo sits in on another Brightwell planning call — the same kind of meeting that nearly scared her off. Bao says:

"The new revision-reminder feature is built, but the backend's hitting the school calendar API too hard and we're seeing bugs in the cloud, so let's not ship to all stakeholders Friday — we'll release to one pilot school, watch the churn metric, and iterate. It's on the roadmap regardless."

In week one, that sentence was a wall. Now Teo unpacks it almost without trying:

• backend hitting the API too hard — the behind-the-scenes engine is asking the school's calendar system for data like a waiter running too many orders to one kitchen.
• bugs in the cloud — things breaking on the remote servers, not on anyone's laptop.
• don't ship to all stakeholders, release to one pilot school, watch the churn metric, iterate — hold the wide launch, go to one school, see if people cancel, improve in small steps.
• on the roadmap regardless — still the plan; this is a when, not an if.

She didn't write a line of code. She followed a sentence that would've been noise a month earlier — and afterward asked Bao one good question: "When you say the API's getting hit too hard, is that the school's system limiting us, or ours?" Bao said "good question, theirs," and later told Renske the career-changer asked sharper questions than some engineers. That's what this vocabulary buys you. Not expertise. Access.