TechPrep LK - AI Career Assistant

Mock Interview
Cv Tailor

Inspiration

Transitioning from university to the professional tech industry can be daunting. In Sri Lanka, securing an internship or associate role at top-tier software companies like Epic Lanka, Rootcode, IFS, or Octave requires rigorous preparation, especially in core concepts like OOP, DSA, and REST APIs. I wanted to create a tool that democratizes interview preparation by simulating real-world technical interviews and automating the tedious process of writing highly targeted cover letters for these specific companies.

What it does

TechPrep LK is a two-fold career preparation assistant: Mock Interview: A chat-based AI interviewer. Users select their target company and job role (e.g., Full-Stack Developer). The AI adopts the persona of a Senior Technical Interviewer from that specific company, asking relevant technical questions, evaluating the user's typed answers, and providing real-time, constructive feedback before moving to the next question.

CV Tailor: A split-screen utility where users paste their current CV content. By selecting a target company and role, the AI instantly generates a highly professional, customized cover letter that highlights the most relevant skills for that exact position.

How we built it

I built the entire application using the MeDo AI platform. First, I used natural language prompts to generate the full-stack architecture, specifying a clean, developer-friendly UI utilizing React and Tailwind CSS with a default dark mode.

Once the UI layout—including the sidebar navigation, dropdowns, and split-screen components—was generated, I integrated MeDo's Large Language Model plugin. I utilized deep-build prompts to wire the UI components to the LLM, creating the logic for the continuous chat history and the cover letter generation without writing manual backend boilerplate.

Challenges we ran into

The primary challenge was prompt engineering for the conversational logic. Getting the AI to wait for the user's response instead of generating the entire interview script at once required careful refinement of the deep-build prompts. Additionally, during development, we experienced temporary 500 Internal Server Errors from the AI API due to high traffic, which required patience and testing to ensure the workflows triggered correctly once the servers stabilized.

Accomplishments that we're proud of

I am incredibly proud of how perfectly the UI mimics a premium SaaS product. Successfully linking a dynamic LLM to a chat interface that maintains conversation history, and tailoring it specifically to Sri Lankan tech giants, was a major win. Achieving all this rapidly through MeDo's visual and prompt-based editor showcased the true power of AI-assisted development.

What we learned

I learned that natural language is becoming the new programming language. Structuring prompts with clear constraints (like specifying the exact tech stack and UI layout) drastically improves the generated output. I also learned how to seamlessly bridge frontend components with LLM actions using MeDo's integrated workflow systems.

What's next for TechPrep LK - AI Career Assistant

In the future, I plan to expand the platform by integrating more local and international companies. I also want to explore connecting local, privacy-focused LLMs (like Ollama running Qwen models) to ensure user data remains completely secure, and add a feature that grades the user's CV against specific job descriptions.