I will make a custom rag application using openai gpt or opensource model
About this gig
Looking to unlock the value of your private data? You're in the right place. I'm Sarab Dar, an AI Developer specializing in Retrieval-Augmented Generation (RAG) and LLM Integrations.
What I Can Do For You:
- Advanced RAG Systems (Chat with Your Data) Connect your PDFs, CSVs, SQL Databases, or internal documentation to GPT or GROQ. I use LangChain and Vector Databases to ensure high-accuracy retrieval with minimal "hallucinations."
- Flexible AI Architectures I offer deployment via Streamlit (for fast, interactive UIs) or FastAPI (for high-performance, scalable backends) to integrate directly with your existing software.
- Open-Source Excellence Worried about API costs? I can work with Open-Source LLMs (Mistral, Llama, Groq) to keep your your overhead low.
What I Need to Get Started:
- Project Goal: A brief description of what you want the AI to achieve.
- Data Source: Your documents or database access (if applicable).
- API Keys: OpenAI / Anthropic / Groq / Azure models
Get to know Sarab Dar
ML Engineer
- FromPakistan
- Avg. response time1 hour
Languages
Urdu, English
My Portfolio
FAQ
Can the chatbot "talk" to specific documents (PDF, CSV, SQL)?
Absolutely! This is the core of a RAG (Retrieval-Augmented Generation) system. I will build a pipeline that "indexes" your private data into a Vector Database so the AI can retrieve and answer questions based solely on your provided information.
What is the benefit of a FastAPI vs. a Streamlit interface?
Streamlit is perfect if you need a beautiful, ready-to-use web dashboard to interact with your AI immediately. FastAPI is best if you are a developer or have an existing website/app and just need a high-performance "brain" (API endpoint) to plug into your current system.
How do i ensure the AI doesn't "hallucinate" or make things up?
I implement advanced RAG techniques, including strict system prompting and "Source Attribution." This means the AI is instructed to only answer based on the provided context. If the answer isn't in your data, the bot will honestly state it doesn't know rather than making up a fact.

