Black Mirror is one of my favorite series. But last weekend, I had a choice: watch the latest Black Mirror series or to explore trending AI tools. Ultimately, I end-up building an AI chatbot trained on my own CV. I learned how AI can turn your resume into a conversational powerhouse. Here’s how I did it but most importantly why you should care.
How it works: No jargon, just results
Let’s break down the tech magic behind the scenes:
- You ask a question Example: “Did Anish receive any awards?” The frontend sends your query to the backend.
- The bot reads my CV (Like a Pro)
- Loads my PDF resume.
- Splits it into bite-sized chunks (because even bots don't like long documents 😁)
- Now we will be diving into specific knowledge about AI.
- The words received will be turned into “Fingerprints”. I am using Cohere’s embeddings.
- Each CV chunk gets a unique code — a fingerprint that helps the AI quickly find relevant info.
- These fingerprints are stored in a vector store (a fancy database for AI brains).
- Matching your question to my CV, your question gets its own fingerprint. The system scans the vector store to find CV chunks with similar fingerprints.
- GROQ’s AI does the heavy lifting, the matching CV chunks + your question go to GROQ’s API. Finally, it generates a human-like answer. Strict rule which I have defined: If the answer isn’t in my CV, it says “No Data Found” (no hallucinated fake skills here!).
- You get the answer The response pops into your chat window. Seamless, fast, and 100% about me.
Lot of how discussed but here is why - some use case scenarios.
- Support-Chatbot: Which you have seen is a very simple example of fetching info from CV. Now consider yourself a brand having product documents exists in the system. and your support-chatbot answers them all—no more “Let me transfer you.”
- Resume-Finder-Chatbot: As an HR - for a job opening you received 100s OR 1000s of resumes. Ask your chatbot to filter down resumes with specific skillset or specific experiences.
- Multilingual Support Bot: If you got some internal WiKi about org, product or policies. use Translation API and create new documents. Which can enable your bot to be multilingual answer to FAQ or queries.
This isn’t Sci-Fi, but still cool enough to fascinate all of us
- No Black Mirror kind of surprises: The bot only answers based on my CV. No existential crises here.
- AI Doesn’t replace humans—It mirrors them: This bot isn’t “fake”; it’s a reflection of my actual experience.
- Keep It Simple: You don’t need a PhD to implement this. Just clear logic and the right tools. I don't have any core-knowledge or expertise about this whole implementation. I understood and implemented via Vibe coding only. (I am sorry, vibe what...!! - about this in next weekend post).
Try it yourself
AI is like a playground. Don't stop yourself to watch video or complete certification. but also build something that solves a problem—even if it’s just your own. There can be still limitation in this implementation and bot. Intention was to show capability and possible reach of this Free AI implementation. Let me know your thoughts in LinkedIn Post comments here.