Description:
An AI chatbot that handles customer support around the clock, such as return requests and complaints, so businesses can answer customers quickly without a human on every chat. Customers ask questions in a simple chat, and the bot replies with clear steps and sends them to the right page when they need to take action (for example, a returns portal). When needed, the AI can trigger real actions, like processing a return request, and send an email confirmation.
How it’s built: A React 18 and TypeScript frontend (Vite), a Python FastAPI backend, and OpenAI GPT-4o-mini with function calling. Login security uses JWT tokens and bcrypt password hashing. Data is stored with SQLModel (SQLite for local development, PostgreSQL for production). Email confirmations are sent from the backend, and the app runs with Docker Compose. English is active and Urdu support is in place.
Who it’s for: Online stores and service businesses that want faster replies and fewer repetitive support tickets.
0
24
Title: Course Companion FTE / StoryForge | AI-Powered Learning Platform
Description:
Built an AI-powered education platform as part of Agent Factory Hackathon IV (Panaversity). It works as a digital tutor that is available 24/7 and supports students and teachers.
Teacher dashboard: Teachers can switch between Student and Teacher views and see class progress at a glance: total students, assigned and completed work, pending items, and average score. A student table shows each learner’s level, stories completed, average score, last active time, and progress bar. The dashboard also includes upcoming deadlines, recent activity, and a tip flagging which student needs extra support.
Learning features: Content delivery, chapter navigation, source-grounded Q&A, rule-based quizzes, progress tracking with streaks and scores, and a free/premium access gate.
AI agent skills: Four skills: concept explainer, quiz master, Socratic tutor, and progress motivator.
Architecture: Built in three phases: a ChatGPT App with a deterministic FastAPI backend (no backend AI calls, to keep costs low), a hybrid phase with selected AI features, and a full Next.js web app with a learning dashboard and analytics.
Process: Followed a spec-first workflow (spec, plan, tasks, implement) with documented decisions and Claude Code.
0
33
Title: Claude Code 101 Certificate | Anthropic
Description:
Completed the Claude Code 101 course from Anthropic, covering how to use Claude Code, an AI coding assistant that works in the terminal, to plan, write, and review code. The course focused on working with an AI agent step by step on real coding tasks.
I use Claude Code in my own projects together with a spec-first workflow (write the spec, plan, tasks, then implement), which helps me build full-stack AI apps faster and keep the code organized and documented. Examples include a bilingual clinic chatbot, a multi-agent travel planner, and an AI accounting assistant.
0
36
Title: Agentic AI Level 1 Developer Certification | PIAIC
Description:
Completed the Agentic AI Level 1 Developer program from the Presidential Initiative for Artificial Intelligence and Computing (PIAIC), a national training initiative in Pakistan. The certificate was awarded on January 30, 2026 (Certificate No. 2026010254199).
The program covered building AI agents that can use tools, follow instructions, and complete tasks step by step. I applied this training in my own projects, including a multi-agent travel planner, a bilingual (English and Urdu) clinic chatbot, and an AI accounting assistant that asks for human confirmation before saving any entry.
Skills / tags: Agentic AI, OpenAI Agents SDK, Python, Tool Calling, Multi-Agent Systems, Chatbots
0
40