Hi there, I'm
I build AI-powered backend systems with Python, LLMs, and RAG pipelines — and craft clean frontends with React & Next.js turning ideas into scalable, production-ready applications.

I'm Asfand Yar, a Gen AI Backend Developer with hands-on industry experience building LLM-powered systems, RAG pipelines, and scalable AI APIs for enterprise applications.
I specialize in Prompt Engineering and Retrieval-Augmented Generation (RAG) — optimizing language model outputs for accuracy, context-awareness, and low-latency production use cases.
With a strong backend foundation in Python, FastAPI, and Node.js, I bridge the gap between cutting-edge AI research and real-world software products.
AI & LLMs
Prompt Engineering, LangChain, OpenAI, RAG
Backend
Python, FastAPI, Node.js, REST APIs
Databases
MongoDB, PostgreSQL, Vector DBs, Redis
Tech Stack
Industry experience building production AI systems and backend services.
Emumba
Developers Hub Corporation
CodeAlpha
A comprehensive toolkit built through real-world projects and continuous learning.
Real-world AI systems, RAG pipelines, and client applications built in production environments.
Conversational voice agent built with RAG and a local LLM served via LM Studio. Processes user speech, retrieves context from a vector store, and generates accurate responses — fully offline, zero cloud dependency.
Prompt engineering framework to benchmark and optimize LLM outputs. Includes automated evaluation pipelines, few-shot template generation, and performance tracking across model versions.
Automated ETL pipeline to extract, clean, and format raw datasets for machine learning model training and benchmarking. Cut data preparation time by over 60%.
Client website for a healthcare practice in Virginia. Clean, accessible design with appointment booking, service listings, and mobile-first layout.
AI-powered web application for clients. Features intelligent automation, a clean user interface, and seamless API integrations for real-world AI use cases.
Intelligent chatbot with natural language processing using NLTK and intent classification, backed by a Flask REST API.
This portfolio built with Next.js, Framer Motion animations, dark/light theme, lazy-loaded sections, and a fully responsive mobile-first design.
Classic Snake game built in Python with Pygame. Features smooth controls, score tracking, increasing speed levels, collision detection, and a retro arcade-style UI.
Dynamic React application for URL manipulation and redirection with real-time preview, input validation, copy-to-clipboard, and a clean responsive UI.
From prompt design to production AI systems — end-to-end Gen AI backend services.
Designing, optimizing, and evaluating prompts for LLMs to maximize accuracy, consistency, and performance across enterprise use cases.
End-to-end Retrieval-Augmented Generation systems with vector search, semantic chunking, and context-aware LLM responses for accurate document Q&A.
Conversational voice agents powered by local or cloud LLMs with RAG integration — fully offline capable, low-latency, and production-ready.
Scalable Python backends with FastAPI and Node.js, integrating LLM capabilities, vector databases, and third-party AI APIs for real-world applications.
Automated ETL pipelines to extract, clean, and format raw data into high-quality datasets for ML training, benchmarking, and AI evaluation.
Seamlessly integrating OpenAI, local LLMs via LM Studio, and open-source models into existing products — with evaluation pipelines and performance tracking.
Responsive, pixel-perfect UIs built with React and Next.js, Tailwind CSS, and Framer Motion — from landing pages to full web applications.
End-to-end websites for clients with clean design, mobile-first layout, SEO optimisation, and fast deployment — from healthcare to SaaS.
Benchmarking open-source and proprietary LLMs, reducing latency, and optimising inference pipelines for production AI systems.
Have a project in mind? Let's build something great together.
Start a Project →