«First Offer» is a Telegram media hub for students in creative fields and junior designers, designed to help them land their first job or design freelance commission. We don’t offer generic advice; instead, we analyze real portfolios and test assignments from practicing designers, publish only junior-level vacancies, and teach our readers how to view their work through an employer’s eyes. Our goal is to shorten the journey from «course completion» to «receiving an offer» through practical experience, constructive feedback, and real-world case studies.

For the «First Offer» media project, we utilize several categories of data, which are divided into internal (community-generated) and external (sourced from public domains).User Data. The primary source is the audience itself. We collect portfolio links, completed assignments, and requests for critiques via Google Forms, a dedicated Telegram bot, and comments on our posts. This data is essential for the «Portfolio Review» and «Weekly Test» segments.Expert Insights. We interview practicing designers, converse with mentors in private chats, and gather their feedback on junior students' work. Based on this data, we create checklists, guides, and portfolio evaluation criteria.Job Openings. We aggregate current listings from HH.ru, LinkedIn, company Telegram channels, and professional chats. The main filter is for Junior roles and internships that do not require «3+ years of experience.» This forms the basis of the «Junior Job Openings» segment.References and Case Studies. For instructional posts (how to / how not to), we draw examples of strong portfolios and successful test projects from Behance, Dribbble, and Figma Community.Market Analytics. For analytical posts (e.g., «Why 70% of juniors fail tests»), we use data from HSE research, HR agency reports, and labor market statistics.

We use a hybrid approach: the machine (automation and AI) handles the routine tasks, while the human (expert and editor) focuses on meaning and quality.

Data Processing Flow:

1. Data Collection (Input):

Machine: Parsing job postings from HH.ru and Telegram channels using keywords («junior, ” „internship, ” „no experience“). Gathering links from Google Forms.

Human: Selecting interesting case studies, communicating with portfolio creators, and sourcing experts for analysis.

2. Preliminary Processing (Filtering):

Machine: Automated job screening (removing requirements like „3+ years of experience, ” checking for scams). Sorting portfolios by specialization (UI/UX, graphic design, illustration).

Human: Vetting job postings for „reasonableness“ (checking for hidden red flags). Filtering out portfolios that do not meet the required level or quality standards.

3. Deep Processing (Analysis):

Machine: AI (such as ChatGPT Vision or specialized APIs) conducts an initial portfolio analysis: „Is there a task description?“, „Is the hierarchy maintained?“, „Is there too many projects?“. Automatically generating tags for the posts.

Human: A practicing designer manually reviews the portfolio and test assignments. Providing subjective expert evaluations—such as, „Why this might not succeed in an interview, ” „How to improve it, ” or „What the employer is looking for.“ Writing the final post copy.

4. Publication (Output):

Machine: Automatic post layout (adding emojis, paragraph breaks), scheduled posting (SMM planner), and cross-posting to TikTok/Shorts.

Human: Final proofreading of the text, approving the visual assets, and responding to subscriber comments.

To ensure automation does not harm media reputation, we implement strict rules for machine processing:1. The «Human-in-the-Loop» Rule:AI may analyze portfolios but must never render a final verdict. AI only highlights areas for growth (e.g., «missing description»). The final critique and recommendations are written solely by a human expert.Automatically generated post text is not published without editorial review.2. Data Collection and Parsing Guidelines:The job vacancy parser must only operate with open APIs or authorized sources (adhering to robots.txt).Subscribers' personal data (links to personal profiles, full names) is not shared with third-party neural networks without explicit user consent. For AI analysis, links are anonymized.3. Vacancy Filtering Guidelines:The automated filter must block vacancies with phrasing like «Junior with 2+ years of experience» and «Intern (work experience is mandatory), ” and delete posts showing signs of fraud (asking for payment for employment, „create 10 mockups for free“).4. Style Guidelines (Tone of Voice):Automated bot responses (if applicable) must be polite but free of a „mentoring“ tone. Using boilerplate phrases like „Your request has been received“ is prohibited. Only use conversational language: „Hi! We got your link, we’ll look at your portfolio soon.“5. Security and Ethics Guidelines:AI must not be used to generate „fake“ reviews or artificially inflate reactions.All portfolio reviews are published only with the author’s consent (this is manually verified).