Afghan Eka Portfolio and Blog
website·Indexed 31 days ago
https://aep.my.idPersonal website of Afghan Eka Pangestu, a software developer specializing in web applications and APIs. Includes a blog featuring technical articles on topics such as the challenges of AI prompting and developer workflows.
What's inside Afghan Eka
- Afghan Eka Pangestu is a Software Developer specializing in Web Application and API development. He focuses on maintainable architecture, scalability, and translating business requirements into technical solutions. He is currently available for freelance opportunities.
Contact Afghan Eka Pangestu for Freelance
For software development, web application, API, or enterprise project opportunities, Afghan Eka Pangestu can be reached via the following channels:Email: afghanekapangestu@gmail.com Phone: 085156283645 LinkedIn: https://www.linkedin.com/in/afghan-eka-pangestu/ GitHub: https://github.com/afg2002 Location: BogorUse an incremental workflow for reliable AI prompting
Rather than attempting to solve a complex problem with one giant prompt, use an iterative process grounded in evidence. This reduces the risk of hallucinations and makes results easier to review.
Incremental workflow steps:
- Request a narrow investigation of the issue.
- Verify the model's hypothesis against the actual codebase.
- Implement the smallest useful change.
- Execute a focused test or build.
- Feed the actual output/error from the test back into the next prompt to refine the solution.
Use an incremental workflow for reliable AI prompting
Avoid using one giant prompt to solve complex problems. Instead, use an incremental workflow grounded in evidence to make the process more reliable and easier to review:
- Narrow Investigation: Ask the AI to investigate a specific part of the problem first.
- Hypothesis Check: Verify the AI's findings against the actual codebase.
- Smallest Change: Request the smallest possible useful change based on the verified hypothesis.
- Focused Testing: Run a specific test or build to verify the change.
- Feedback Loop: Feed the actual result (including failing tests or error logs) back into the next prompt to refine the solution.
Apply a technical specification framework to AI prompts
To avoid vague outputs and incorrect assumptions, treat prompts as small technical specifications. Instead of general requests, define these five elements to ensure the model has the necessary information to produce a useful and consistent result:
- Goal: Define the specific behavior that needs to change.
- Context: Provide relevant files, existing code patterns, and specific error messages.
- Constraints: Specify what must remain unchanged and which tools or libraries are permitted.
- Output: Define the exact required format (e.g., a patch, a detailed explanation, or a test case).
- Validation: Provide the specific command or behavior that proves the solution works.
Structure AI prompts as technical specifications
To avoid vague outputs and incorrect assumptions, treat a prompt like a small technical specification. Before requesting a solution, define these five elements to provide the model with necessary context and constraints:
- Goal: The specific behavior you want to change.
- Context: Relevant files, existing patterns, and error messages.
- Constraints: What must remain unchanged and which tools or libraries are permitted.
- Output: The exact format required (e.g., a patch, a detailed explanation, or a test case).
- Validation: The specific command or behavior that proves the result works.
// Instead of: "Fix my API" // Use: Goal: Fix the [Name] endpoint which is returning a 500 error. Context: [Insert response body] and [Insert relevant handler code]. Constraints: The public API signature must remain compatible. Output: A focused code change and a corresponding test case. Validation: Run `npm test tests/api/endpoint.test.ts` to verify the fix.Featured Software Projects
Key projects developed by Afghan Eka Pangestu include:
- SehatSegar.my.id: A Laravel-based POS web application for inventory, cashier transactions, and sales reporting.
- Movie Ticket Booking Web: A booking application built with Vue.js 2, Vuetify, and MongoDB.
- Web Zero Waste Massive: An API for a cleaning service booking application using React and Express.
- Project Management Application: A Laravel application featuring budgeting and vendor management.
- Router Dashboard Application: A monitoring tool built with Spring Boot and Vue.js 2.
- Baca Komik: A Python-based comic scraping and PDF conversion tool.
Technical Skills and Stack
Afghan Eka Pangestu's technical expertise includes the following stack:{ "frontend": ["HTML", "CSS", "JavaScript", "Vue.js", "Vue.js 2", "Vuetify", "React"], "backend": ["PHP", "Laravel", "Java", "Spring Boot", "Golang", "Express"], "database_and_tools": ["MongoDB", "MySQL", "Git", "Terminal", "VS Code"], "professional_strengths": ["Clean Architecture", "Scalability", "Maintainability", "Teamwork", "Communication", "Problem Solving", "Adaptation"] }Software Development Experience
Professional experience includes roles at the following organizations:
- Generali Indonesia: Java Developer focusing on insurance business applications and backend services.
- PT Inmotion Inovasi Teknologi: Web Developer specializing in chatbots, omni-channel solutions, Generative AI, and Salesforce integration.
- PT BTPN Syariah Tbk: Web Developer (Intern) building router applications for dashboard listing and log monitoring using Spring Boot and Vue.js 2.
- PT Ciptanusa Nuansa Mulia: Freelance Web Developer creating project management applications with budgeting (RAB) and vendor management using Laravel.
- PT Bintang Bahagia Selalu — Royal Star: Freelance Web Developer improving UI/UX and frontend components.
- CV Mitrasoft Global: Intern Web Developer migrating desktop back-office applications to web platforms using Java Spring.