Agentic AI Engineer · FastAPI & Python

FastAPILangGraphNext.jsPostgreSQL

|

Building systems where AI agents, APIs, and databases talk to each other in production.

Full-stack engineer specializing in multi-agent AI systems, FastAPI backends, and Next.js frontends — with two platforms live in production, serving real users today.

View Projects Download Resume

GET /about

I didn't start in tech.

I started in digital marketing and sales — reading people, selling ideas, closing deals. Then I taught myself to build the systems I used to just sell. Now I ship production software: FastAPI backends, LangGraph multi-agent pipelines, and Next.js frontends that real people use every day.

Two platforms I've built are live right now — one serving 1,400+ active gym members, another managing real people's personal finances with an AI advisor built in. I care less about how a system looks in a demo and more about whether it holds up when someone depends on it.

Currently finishing my CS degree at Superior University, Lahore — and building the next thing.

1,400+

Active users on live systems

Production portfolio projects

2026

AI Engineering certified (Datics-AI)

GET /skills

AI / Agents

LangChain
LangGraph
RAG · FAISS
Prompt Engineering
OpenAI · OpenRouter

Backend

Python
FastAPI
PostgreSQL
JWT Auth
SQLAlchemy · REST

Frontend

Next.js
React
TypeScript
Tailwind CSS

Tools

Git / GitHub
AWS · Azure
Vercel · Render
Docker
Cursor · Antigravity

GET /projects

Two things I've shipped and kept alive in production, plus a few I've built to learn the edges of what's possible.

FEATURED · 2026

IntelliSpend

AI-powered personal finance platform

The problem: most budgeting apps show you a number and hope you trust it. IntelliSpend computes every balance live, from the transaction ledger, every time — nothing cached, nothing that can silently drift from reality.

What it does

✓Multi-ledger accounting across accounts, transactions, transfers
✓Real-time computed balances (never stored/cached)
✓AI financial advisor — grounded in live account data, 30-day transaction history, and pending bills
✓Recurring bills & reminder engine
✓JWT-authenticated, responsive dashboard

Tech stack

Next.jsFastAPIPostgreSQL (Neon)JWT AuthOpenRouter (GPT-4o-mini)

Architecture

User → Next.js Frontend → FastAPI REST API → PostgreSQL (Neon)

OpenRouter (GPT-4o-mini) — AI Advisor

The hard part

Keeping balances accurate across multiple ledgers without caching meant every dashboard load recomputes from the raw transaction log — fast enough to feel instant, correct enough to trust. The AI advisor needed the same discipline: it only answers from data injected at request time — live balances, recent transactions, pending bills — never from memory, so it can't hallucinate a number that isn't real.

Live Demo →GitHub →Watch 2-min walkthrough →

CLIENT PROJECT · 2025Confidential visuals

Gym Management Platform

Production backend serving 1,400+ active members

Built and deployed the FastAPI backend for a gym operator managing subscriptions, membership expiry, and biometric access control at scale.

Client project — architecture & challenge only

Screenshots and UI details are omitted to protect client confidentiality. The write-up focuses on backend architecture and technical problem-solving.

What it does

✓Subscription & membership expiry tracking
✓Biometric access control (ZKTeco K50 / pyzk)
✓Automated template backup/restore on renewal
✓End-to-end ownership — requirements through deployment

Tech stack

FastAPIPostgreSQLpyzkJWT Auth

The hard part

Integrating a ZKTeco K50 biometric access device via the pyzk library — the challenge wasn't the integration itself, it was handling membership expiry without forcing members to re-enroll their fingerprint every renewal. I built a template backup/restore pattern: biometric templates are preserved server-side and re-synced to the device automatically on renewal, so access just resumes.

Multi-Agent AI Assistant System

Python · LangGraph · LangChain · Tavily

Three specialized agents orchestrated via LangGraph, with real-time web search through Tavily and a Streamlit frontend for multi-agent interaction.

GitHub →

ExamPro

Python · RAG · Embeddings

RAG pipeline that pulls exact, context-aware answers out of PDFs — built for exam prep, not vague summaries.

GitHub →

LinkedIn Job Extractor

Python · Apify · LangGraph · Next.js

Scrapes listings, filters them against your criteria with AI, and auto-sends cover letters through an email API.

GitHub →

GET /experience

2025 — Present

Freelance — AI & Backend Developer

Lahore, PK

Independent client engagements, owned end-to-end: requirements gathering, iterative delivery, production deployment.

Shipped a FastAPI backend now running in production for a gym platform serving 1,400+ active members — subscriptions, membership tracking, and biometric access control.

BS Computer Science

Superior University, Lahore · CGPA 3.3/4.0

2023–Present

AI Engineering Certification

Datics-AI (NAVTTAC)

POST /contact

Open to freelance work and full-time AI/backend roles.

Email LinkedIn GitHub

email: abdulsabooriftikhar3718@gmail.com