Skip to content
Ahmed Raza — AI Engineer, KarachiAvailable for work

I build AI productsthat actually ship.

Complete web applications and the production AI inside them — RAG pipelines, tool-using agents and eval harnesses, from first commit to live deployment.

Selected work ↓Résumé ↗
06
Products live
01
PyPI package
78
Tests, CI-gated
’26
BS AI, FAST-NUCES
Profile

I'm an AI Engineer with a BS in Artificial Intelligence from FAST-NUCES.

I build complete web applications with Next.js, React and FastAPI, and I ship the AI features that make them useful — RAG pipelines, chatbots and tool-using agents powered by LLM APIs and local LLMs.

My work runs in production: I deploy with Docker, AWS and Vercel, wire up CI/CD with GitHub Actions, and monitor model quality with LangSmith. Multiple products I've built are live and in use today.

Location
Karachi, Pakistan
Education
BS in Artificial Intelligence — FAST-NUCES, 2022 – 2026
Focus
LLMs & shipping real AI products
Availability
Currently open to full-time AI engineering roles — remote or Karachi.
Ahmed Raza — portrait
Ahmed RazaKarachi · 24.86°N 67.00°E

What I do

Full-stack web development

Complete applications with Next.js, React and FastAPI — REST APIs, auth, databases and clean, responsive UIs.

AI chatbots & RAG systems

Chatbots grounded in your documents — chunking, embeddings, vector search and retrieval tuning for sourced answers.

AI agent development

Tool-using agents built with LangChain and LangGraph that execute multi-step tasks via APIs and webhooks.

Cloud deployment & DevOps

Production deployments with Docker, AWS and Vercel — Terraform provisioning, CI/CD and LLM observability.

Stack

Tools I ship with

Frontend & Backend

Next.jsReact.jsFastAPIFlaskNode.jsExpressREST APIsWebhooksJWT Auth

Generative AI

LLM APIs (OpenAI, Claude, Gemini)Prompt EngineeringLangChainLangGraphRAG PipelinesAI AgentsTool Use / Function CallingLLM Evals (LLM-as-Judge)CI Quality GatesChromaQdrantFAISSLangSmithLangfuse

Cloud, DevOps & Data

DockerAWS (EC2, S3, Lambda)VercelTerraformGitHub ActionsLinuxGitClaude CodePostgreSQLSupabaseMongoDBMySQL

Languages & ML

PythonTypeScriptJavaScriptSQLpandasNumPyPyTorchTensorFlowscikit-learnKeras

Experience

Where I've worked

  1. Jul 2025 — Jun 2026

    Karachi, Pakistan

    AI Engineer — HashOne Global

    • Built AI applications and chatbots end to end using LLM APIs and local LLMs, applying prompt engineering to produce clean, structured output.
    • Designed RAG pipelines over vector databases — document chunking, embedding and retrieval tuning — to ground answers in source documents.
    • Developed AI agents with LangChain and LangGraph that use tool calling to execute multi-step tasks, served through FastAPI REST APIs with webhook integrations.
    • Deployed services with Docker and CI/CD pipelines on AWS, and monitored model quality with LangSmith.

  2. Jun 2024 — Aug 2024

    AI Engineer Intern — Tech Hunt

    • Built RAG and document-processing pipelines in Python with LangChain and the OpenAI API.
    • Chunked and embedded documents into a FAISS vector store and tuned retrieval for grounded answers.
Selected work

Built, tested, shipped

2025 · LLM Evals

NovaNet Eval Harness

AI support agent with production-grade LLM evals

A customer-support RAG agent wrapped in a full evaluation pipeline — 100-case golden dataset, deterministic assertions plus LLM-as-judge grading, hash-keyed response caching, and a GitHub Actions quality gate that blocks bad prompt changes before merge. Baseline evals caught an intermittent prompt-injection vulnerability. Ships with a Next.js metrics dashboard: run trends, per-case judge verdicts, grounded chat and a rendered knowledge base.

100-case golden dataset, CI-gated

FastAPI / LangGraph / Groq Llama 3.3 / Chroma / LLM-as-Judge / GitHub Actions / Next.js

Live ↗Code ↗

2025 · Open Source · MCP

PSX MCP Server

Published MCP server for LLM agents

An open-source Model Context Protocol server that gives Claude and other LLM agents live Pakistan Stock Exchange data — quotes, indices (KSE-100), company fundamentals, dividends and announcements — as 10 tool-callable endpoints. Built test-first with 78 tests, CI across Python 3.10–3.13 on Linux and Windows, and published to PyPI for one-line install (uvx psx-mcp-server).

78 tests · live on PyPI

Python / MCP (FastMCP) / httpx / BeautifulSoup / pytest / GitHub Actions / PyPI

PyPI ↗Code ↗

2025–26 · Voice AI · FYP

Convey AI

Conversational AI interview agent

An AI agent that conducts and evaluates mock interviews autonomously in Urdu and English, with an avatar-based interview mode. Built as my Final Year Project — deployed and live.

Interviews in Urdu & English

Next.js / Node.js / OpenAI API / REST APIs

Live ↗Code ↗

2025 · AI Agents · RAG

Cognos

Full-stack AI agent app (RAG + tools)

Chat with a tool-using AI agent over your own documents, with built-in web search. JWT auth, per-user Chroma vector stores, Postgres-backed chat memory and token-by-token SSE streaming. Deployed on AWS EC2, provisioned with Terraform.

Per-user vector stores · SSE streaming

FastAPI / LangChain / LangGraph / Chroma / PostgreSQL / Next.js / SSE / Terraform / AWS EC2

Live ↗Code ↗

2024 · Full-stack

CareShare

Medicine donation platform (MERN)

Connects medicine donors with people in need. Includes an admin panel that moderates donations and user accounts before listings go live. Deployed and accessible online.

React / Node.js / Express / MongoDB

Live ↗Code ↗

2024 · Internal tool

IT Inventory Management System

Asset tracking for company IT

Internal tool for tracking company IT assets and maintenance, with a React dashboard on a FastAPI and MySQL backend. Deployed and accessible online.

FastAPI / MySQL / React.js / REST APIs

Live ↗Code ↗
Private client project

Python · LangChain · OpenAI · FAISS · FastAPI

2024 · Client work

RAG Document Parsing Chatbot

Grounded Q&A over long documents

A chatbot that answers questions over long documents — embeds the text into a FAISS vector store, retrieves the closest matches, and serves grounded answers through a FastAPI service.

Python / LangChain / OpenAI / FAISS / FastAPI

Contact

Have something worth shipping?

I usually reply within a day. Tell me about the product, the problem, or the role — email works best.

ahmedrz.amjad@gmail.com

Currently open to full-time AI engineering roles — remote or Karachi.