How developers can automate the job search: alerts, a tracker, AI and interview prep

A job search turns into chaos fast: twenty tabs of job posts, three versions of your CV, LinkedIn messages you forgot to answer, and an interview you remember an hour before it starts. Automation here isn't about a robot applying for you. It's about jobs coming to you, nothing getting lost, and your time going to what actually moves the result: good applications and preparation.

The workflow has five parts:

1. Jobs come to you (alerts).
2. Every application is logged (tracker).
3. CV and cover letter get tailored fast (AI, under your control).
4. Messages to people have a rhythm (outreach).
5. Interviews are prepared to a plan (prep loop).

1. Alerts: jobs come to you

Set it up once in an hour, then just read fresh jobs once a day.

LinkedIn. Job search accepts boolean: quotes for phrases, AND, OR, NOT and brackets.

Filters: location or Remote, posted in the past week, then Set alert with daily frequency.

Google Alerts on ATS domains. Many companies post on Greenhouse, Lever or Ashby before jobs reach aggregators. An alert catches them directly:

Djinni. A saved search arrives by email or Telegram. If your technology rarely appears in job titles, switch to full-text search, or the subscription stays empty.

DOU. RSS feeds by search and category, for example https://jobs.dou.ua/vacancies/feeds/?category=Front%20End. To search job descriptions, add &descr=1. Any RSS reader works (Feedly, Inoreader, Miniflux).

Work.ua. The results page has a button to get new jobs in Telegram or Viber.

Remote boards. Remote OK has RSS for every tech tag. Remotive, Himalayas and Wellfound have email alerts.

The full list of sites: Where developers find jobs.

2. Tracker: nothing gets lost

Past ten applications, you'll start mixing up companies without a tracker. Two options.

Ready-made trackers. Huntr and Teal save a job in one click through a browser extension, keep a status board and compare your CV with the job post. Jobscan focuses on keyword match. All have a free plan; paid plans change often, so check their sites for prices.

A spreadsheet. Notion, Google Sheets or Airtable. Free and flexible. Minimum columns:

Once a week, look at the numbers the way salespeople look at campaigns: applications, replies, interviews. The ratios tell you where the problem is:

• Few replies to applications. The CV, or what you're applying for. Check fit, keywords, level.
• Screens but no technical rounds. How you talk about yourself and your salary expectation.
• Technical rounds but no offers. Interview prep. Go to part 5.

3. AI for CVs and cover letters

AI is good at three things: pulling requirements out of a job post, showing what your CV is missing, and rewriting text for a specific role. It's bad at one: sticking to the truth. So the rule is simple.

AI only rewrites what's already in your facts. No skills, projects or numbers that weren't there.

Workflow:

1. Build a base CV with every fact: projects, numbers, stack. It's longer than what you send and never goes out.
2. Paste the job post into an AI assistant and ask it to extract: role, level, required skills, nice-to-haves, key phrases verbatim.
3. Ask it to compare with your base CV and list the missing keywords, without rewriting.
4. Decide yourself what's true. Then ask it to rewrite the summary and 3-5 experience bullets for the role, using only facts from the base CV.
5. Read it all out loud. If a phrase doesn't sound like you, rewrite it.

Example prompt:

Here's a job post and my base CV. List the job's required qualifications. Then show which ones my CV has, which it lacks, and which it has under a different name. Don't add anything of your own.

Cover letter. 4-6 sentences: why this company (one specific fact about it), what you have that they need (1-2 results), and what you suggest next. AI gives a good draft, but write the first line yourself; that's where generated text shows.

Privacy. Before pasting your CV into a public chatbot, remove your phone, email and address. AI doesn't need them.

Vacancy mode in ShortLesson

We're building this workflow into ShortLesson as vacancy mode. You paste a job post and get:

• a match score with a visible formula, not an "AI rating" that changes from run to run;
• missing keywords, with suggested edits you accept or reject;
• a cover letter in Ukrainian or English, from your facts only;
• interview questions for that exact job and lessons that close the gaps;
• a prep plan up to the interview date.

While vacancy mode is in progress, you can already practise the technical part in interview practice.

4. LinkedIn outreach rhythm

Messages to recruiters and hiring managers often get more than an application through the site. Message templates are in the LinkedIn guide. This part is about rhythm.

• 15-20 minutes a day, not three hours once a week. Consistency works for your profile and for you.
• 5-10 new contacts a day at companies you applied to or want to. Before the request: view their profile and react to a recent post.
• Follow-ups: first after 5-7 days, second a week later, then stop.
• 2 posts or 10 thoughtful comments a week in your niche, so the profile people open after your request looks alive.
• Log everything in the tracker: who, when, what they said.

Salespeople, for whom this is the whole job, treat an acceptance rate below one in five as a warning sign. It's a useful benchmark for job search too: if fewer accept, fix your profile or who you write to.

5. Interview prep loop

Preparation is a process you can put on rails too.

Before each interview:

1. Reread the job post and list the 5-7 topics they'll surely ask about.
2. Work through questions on those topics. In ShortLesson these are scored 0 to 3 and repeat on a schedule, so weak topics come back until they stick.
3. Prepare a 90-120 second introduction: who you are, what you work with, 2-3 facts with numbers, why this role. Say it out loud several times, ideally on camera.
4. Prepare 8-10 STAR stories (situation, task, action, result) for behavioural questions: a conflict, a mistake, a tough deadline, a decision you're proud of. For English interviews, write them in your language first, then translate and say them aloud.
5. Prepare 3 questions for the company.

After each interview (same day):

1. Write every question you remember into the tracker.
2. Mark the ones you answered poorly.
3. Add them to your review.

After a few interviews you'll have a personal question bank, and each interview gets easier than the last.

To practise English under pressure, try free peer mocks on Exponent and AI mocks on interviewing.io.

Why not to switch on auto-apply bots

Some services promise to apply to hundreds of jobs for you. Sounds handy, but:

• Platform rules. LinkedIn explicitly prohibits third-party software and extensions that act on your behalf, and can restrict or close your account. In a job search, losing your LinkedIn account costs more than any number of applications.
• Quality. Mass applications land on jobs that don't fit. Recruiters see identical, untailored applications, and they remember your name for the wrong reason.
• Unknown forms. The bot answers form questions for you, sometimes wrongly: visa, salary, experience.
• The numbers lie. 300 applications and 2 replies don't tell you what's wrong.

Automate what surrounds the application: finding jobs, reminders, tracking, drafts. Keep the application itself, the choice of job and the final text for yourself.

Weekly plan