The latest articles on DEV Community by Elyass (@elyass_43b15fee1a28f165db). https://dev.to/elyass_43b15fee1a28f165db https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4120496%2F6b2b5fdf-7a7a-4a83-a942-14293cd6abce.png https://dev.to/elyass_43b15fee1a28f165db en Elyass Fri, 11 Sep 2026 08:07:01 +0000 https://dev.to/elyass_43b15fee1a28f165db/how-i-built-a-pdf-chat-api-in-one-day-with-fastapi-gemini-and-qdrant-1j94 https://dev.to/elyass_43b15fee1a28f165db/how-i-built-a-pdf-chat-api-in-one-day-with-fastapi-gemini-and-qdrant-1j94 <p>Have you ever wanted to just <em>talk</em> to a PDF instead of reading through 50 pages?</p> <p>I built a full PDF Chat API in one day — upload any PDF, ask questions in natural language, and get AI-powered answers. Here's how I did it.</p> <h2> What it does </h2> <ul> <li>Upload any PDF document</li> <li>Ask questions about its content in natural language</li> <li>Get accurate answers powered by RAG (Retrieval Augmented Generation)</li> <li>Clean web UI included — no frontend framework needed</li> <li>REST API with authentication for easy integration</li> </ul> <h2> Tech Stack </h2> <ul> <li> <strong>FastAPI</strong> — REST API backend</li> <li> <strong>Google Gemini</strong> — embeddings (<code>gemini-embedding-001</code>) + chat (<code>gemini-2.5-flash</code>)</li> <li> <strong>Qdrant</strong> — vector database for semantic search</li> <li> <strong>LangChain</strong> — RAG pipeline orchestration</li> <li> <strong>Pure HTML/CSS</strong> — no React, no framework</li> </ul> <h2> How it works </h2> <p>The architecture is classic RAG in two phases:</p> <p><strong>Ingestion (upload):</strong></p> <ol> <li>Extract text from PDF</li> <li>Split into chunks (1000 chars, 200 overlap)</li> <li>Generate embeddings with Gemini</li> <li>Store in Qdrant</li> </ol> <p><strong>Query (chat):</strong></p> <ol> <li>Embed the user's question</li> <li>Search Qdrant for the 4 most relevant chunks</li> <li>Send chunks + question to Gemini</li> <li>Return the answer</li> </ol> <h2> The code </h2> <p>The core is surprisingly simple:<br> </p> <div class="highlight js-code-highlight"> <pre class="highlight python"><code><span class="n">chain</span> <span class="o">=</span> <span class="p">(</span> <span class="p">{</span><span class="sh">"</span><span class="s">context</span><span class="sh">"</span><span class="p">:</span> <span class="n">vector_store</span><span class="p">.</span><span class="nf">as_retriever</span><span class="p">(</span><span class="n">search_kwargs</span><span class="o">=</span><span class="p">{</span><span class="sh">"</span><span class="s">k</span><span class="sh">"</span><span class="p">:</span> <span class="mi">4</span><span class="p">}),</span> <span class="sh">"</span><span class="s">question</span><span class="sh">"</span><span class="p">:</span> <span class="nc">RunnablePassthrough</span><span class="p">()}</span> <span class="o">|</span> <span class="n">prompt</span> <span class="o">|</span> <span class="n">llm</span> <span class="o">|</span> <span class="nc">StrOutputParser</span><span class="p">()</span> <span class="p">)</span> </code></pre> </div> <p>That's the entire RAG chain — retrieve relevant context, inject into prompt, generate answer.</p> <h2> What I learned </h2> <ul> <li>Gemini embeddings produce 3072-dimensional vectors (not 768 like older models)</li> <li> <code>grpcio</code> on Windows can be a pain — pin to version 1.62.2</li> <li>Qdrant Cloud free tier is genuinely useful for side projects</li> <li>Building a clean UI in pure HTML/CSS takes less time than setting up React</li> </ul> <h2> Try it yourself </h2> <p><a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fjll3vttcnu9fiuq7dfor.png" class="article-body-image-wrapper"><img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fjll3vttcnu9fiuq7dfor.png" alt=" " width="800" height="380"></a> get the packaged version ready to deploy:<br> 👉 <a href="https://elyassdigital.gumroad.com/l/zcgjmmz" rel="noopener noreferrer">https://elyassdigital.gumroad.com/l/zcgjmmz</a></p> <h2> What's next </h2> <ul> <li>Multi-user support with separate collections per user</li> <li>Docker deployment guide</li> <li>Support for other document types (Word, Excel)</li> </ul> <p>Built this as a side project — happy to answer questions in the comments!</p> ai rag python fastapi