WO

Grant Monitoring and Evaluation Officer

Worldvision

Full time Kegonga/ Ntimaru, Kenya, KE
Posted 1 week, 1 day ago 93 views 0 applications

Job Description

Here's a thinking process: 1. **Analyze User Input:** - User is acting as a job listing editor. - I receive raw scraped text from a job website. - Task: Extract ONLY actual job information, return clean, readable format. - Must REMOVE: social sharing, author bylines, "you may also like", navigation, comments, ads/promotions, cookie/subscription notices, duplicate text, category labels not part of job, stray numbers/formatting artifacts. - Must KEEP and STRUCTURE: job title, company, location, job description/responsibilities, requirements/qualifications, how to apply, salary (if mentioned), deadline/closing date, job ID/reference number, contract type/level. - Return ONLY the cleaned job text. No commentary, headers like "Here is the cleaned version", no markdown. Wait, the actual prompt I need to respond to is a bit mixed. Let me re-read carefully. The user says: "You receive a plain-text job description. Your ONLY job is to format it as clean semantic HTML." Then there are rules: - Use

for intro paragraphs and general text - Use

for section headers - Use
  • for list items - No CSS classes, inline styles,
    ,
    - Preserve ALL original content word-for-word — do not add, remove, summarise, or rephrase - If text has no clear sections, just wrap paragraphs in

    tags - Return ONLY the HTML. No markdown fences, no explanation, no preamble. Then there's a "Job description:" section with a lot of text. So the task is: Take the provided job description text, clean it according to the rules (but the rules here are about HTML formatting, not the earlier "remove social sharing etc." - wait, the very first part describes a different role "job listing editor" task, but then the actual prompt says "You receive a plain-text job description. Your ONLY job is to format it as clean semantic HTML." and gives HTML formatting rules. The initial part about removing social sharing etc. might be context from a system prompt, but the actual instruction is the HTML formatting one. I need to follow the most recent/active instructions.) Let me re-read the whole user message carefully: "You are a job listing editor. You receive raw scraped text from a job website. Your job is to extract ONLY the actual job information and return it in a clean, readable format. REMOVE all of the following: - Social sharing text (Share, Facebook, Twitter, WhatsApp, Telegram, etc.) - Author bylines (by AuthorName, written by, posted by) - "You may also like" / related posts sections - Navigation text (Previous, Next, Read more, Back to) - Comments sections - Advertisements or promotions - Cookie/subscription notices - Duplicate text (if the title appears multiple times, keep it once) - Category labels that aren't part of the job (e.g., "Internships" as a standalone line) - Stray numbers or formatting artifacts KEEP and STRUCTURE: - Job title - Company/Organisation name - Location - Job description and responsibilities - Requirements/qualifications - How to apply / application instructions - Salary (if mentioned) - Deadline / closing date - Job ID / reference number - Contract type / level Return ONLY the cleaned job text. Do not add any commentary, headers like "Here is the cleaned version", or markdown formatting. Just the clean job listing text." Then: "You receive a plain-text job description. Your ONLY job is to format it as clean semantic HTML. Rules: - Use

    for introductory paragraphs and general text - Use

    for section headers (e.g. Responsibilities, Qualifications, Requirements, Education, Experience, Skills, How to Apply, Key Duties, About the Role) - Use
    • for list items (duties, requirements, qualifications, skills) - No CSS classes, no inline styles, no
      s, no
      tags - Preserve ALL original content word-for-word — do not add, remove, summarise, or rephrase - If text has no clear sections, just wrap paragraphs in

      tags - Return ONLY the HTML. No markdown fences, no explanation, no preamble. Job description: [long text]" Okay, there are two slightly different instruction sets. The first one is about extracting and cleaning job info, removing fluff, keeping specific fields. The second is about formatting as clean semantic HTML, preserving all content word-for-word, using specific tags. Which one to follow? The user message starts with "You are a job listing editor..." then gives extraction rules, then says "Return ONLY the cleaned job text." Then it says "You receive a plain-text job description. Your ONLY job is to format it as clean semantic HTML." and gives HTML formatting rules, and then provides the job description text. I think the actual task I need to perform is the HTML formatting one, because it includes the actual job description text and specific HTML rules. The first part might be a system prompt or context, but the concrete prompt is the HTML formatting one. However, both are in the same message. I need to reconcile. Let me read the very end: "Job description: [text]" and before that "Rules: ..." The HTML formatting rules say "Preserve ALL original content word-for-word — do not add, remove, summarise, or rephrase". The extraction rules say "REMOVE all of the following: ..." which includes removing social sharing, bylines, etc. The HTML rules don't explicitly say to remove those, but they say preserve all original content word-for-word. There's a conflict. Actually, the HTML rules say "If text has no clear sections, just wrap paragraphs in

      tags" and list specific tags. They don't mention removing anything. The extraction rules explicitly say to remove many things. The job description text included has things like "

      ", "Employee Contract Type:", "World Vision is a Christian humanitarian organization...", "World Vision Kenya:" disclaimer, etc. I need to decide: Should I first clean the text per the extraction rules (remove social sharing, bylines, etc.), then format as HTML per the HTML rules? Or should I just format the given text as HTML, preserving everything word-for-word? The user message structure: It starts with "You are a job listing editor..." (extraction rules), then "You receive a plain-text job description. Your ONLY job is to format it as clean semantic HTML." (HTML rules), then "Job description: [text]". It's possible the user wants me to do both: first clean the text (remove the stuff listed), then format the remaining content as HTML. But the HTML rules say "Preserve ALL original content word-for-word — do not add, remove, summarise, or rephrase". That contradicts removing things. Maybe the intent is: The extraction rules are the general job of the persona, but for this specific task, I'm to format the given text as HTML, and the "preserve all original content word-for-word" means don't alter the text content, just wrap it. But the extraction rules list things to remove. Which takes precedence? Let me look at the actual text provided. It has: - `

      With 75 years of experience...

      ` - content - `

      ` - empty paragraph - `Local - Fixed Term Employee (Fixed Term)` - contract type - `

      Job Description:

      ` - heading - Job description paragraphs - ` ` - italic bold with non-breaking space - "Please Note:" section about Christian mission, devotions, etc. - `

      Job Responsibilities:

      ` and ul/li lists - Many more ul/li lists for responsibilities, qualifications, education, World Vision Kenya disclaimer, etc. - `

      World Vision Kenya:

      ` and ul/li about recruitment - `

      World Vision Kenya is part of the Inter-Agency Scheme...

      ` - `

      Applicant Types Accepted:

      Local Applicants Only` The HTML rules say use

      for section headers,
      • for lists,

        for intro/general text. No
        , no

        , no inline CSS. Preserve all content word-for-word. If I just wrap everything as HTML per the rules, I need to keep all text exactly as is, just change tags. But there are `
        ` tags in the original? Actually the text has `

        `, `

        `, etc. The rules say "No ... no
        tags". So I need to handle line breaks without
        . Probably wrap in

        or just let paragraphs flow. Also, the extraction rules say remove things like social sharing, bylines, etc. The text doesn't seem to have those explicitly, but has the "World Vision Kenya:" disclaimer and "Please Note:" which might be considered part of the job or not. The extraction rules also say "Remove duplicate text (if the title appears multiple times, keep it once)" - not sure about title duplication. Given the conflict, I need to interpret the prompt as a whole. The user says: "You receive a plain-text job description. Your ONLY job is to format it as clean semantic HTML." Then gives rules. Then gives the job description. The HTML rules are the specific instructions for this task. The extraction rules might be background/context from the system, but the actual command is the HTML formatting one. However, the extraction rules explicitly say "REMOVE all of the following:" and list things. If I follow the extraction rules, I'd remove parts of the job description that might be considered essential (like the "Please Note:" Christian mission statement, the disclaimer, etc.). But the HTML rules say "Preserve ALL original content word-for-word". Maybe the correct approach is: First, apply the extraction rules to remove the listed fluff, then format the remaining clean text as HTML per the HTML rules. But the HTML rules say "Preserve ALL original content word-for-word — do not add, remove, summarise, or rephrase". If I remove things first, I'm not preserving all original content of the given text, but I'm following the extraction rules which are part of the persona's job. Let me re-read the very beginning: "You are a job listing editor. You receive raw scraped text from a job website. Your job is to extract ONLY the actual job information and return it in a clean, readable format." Then extraction rules. Then "Return ONLY the cleaned job text. Do not add any commentary, headers like "Here is the cleaned version", or markdown formatting. Just the clean job listing text." Then: "You receive a plain-text job description. Your ONLY job is to format it as clean semantic HTML." This seems like a separate instruction or a reformulation. Maybe the user is giving me two prompts in one, or the second is the actual task format. Given the ambiguity, I should look at the job description text and see what makes sense. The text is a job listing for an M&E Officer at World Vision Kenya. It has a lot of content. The extraction rules would remove things like the "Share" etc., but those aren't present. It has the "Please Note:" about Christian mission, which might be considered part of the job or not. The "World Vision Kenya:" disclaimer is standard but might be removed per "category labels that aren't part of the job". The "Applicant Types Accepted: Local Applicants Only" might be kept or removed. The HTML formatting rules: "Use

        for introductory paragraphs and general text", "Use

        for section headers", "Use

MJC
ECHO
Your MJC Assistant

I'm ECHO, your MJC career assistant. I can help you find jobs, explore career tools, and connect with opportunities across Africa.

How was your experience with ECHO?