Muhammad Abdullah

Full Stack AI Developer specializing in production-ready deep learning and scalable web systems.

Based in: Lahore, PakistanCurrently: Full Stack AI Developer, Seamry SolutionsExperience: 1 year

Muhammad is an AI/ML and full-stack developer with extensive experience delivering production-ready deep learning pipelines and React applications across multiple industry internships. He has a proven track record of building end-to-end intelligent systems, including AI-driven emergency response platforms and high-accuracy NLP models.

Experience

Full Stack AI Developer

Seamry SolutionsLahore, PakistanworkJun 2026 – Present

Summary

Develop and integrate AI/ML-powered features into full-stack web applications, working across model integration, backend services, and UI delivery. Build and maintain production-grade systems spanning frontend, backend, and machine learning components within an end-to-end development workflow

What I did

Separated cardio-related doctors from general data and identified their publication sources using keyword-specific factors.
Working on voice models and conducting discovery for tools like Retell, LiveKit, or ElevenLabs.

Results

Developed a filtering pipeline to categorize medical professionals and their publications based on specific keyword criteria.
Built an in-house extension in two days using LLMs to predict actions for Upwork bids and provide graphical mapping of bidding activity.
Processed almost 50k practitioners with 90% accuracy.

AI/MLFull-stackBackend servicesUI deliveryMachine learningVoice modelsRetellLiveKitElevenLabsLLMs

Computer Vision & Deep Learning Intern

ML-BenchLahore, PakistaninternshipJun 2025 – Aug 2025

Engineered deep learning pipelines for image analysis using CNNs, OpenCV, and TensorFlow on 10,000+ image datasets. Improved model accuracy by 15% and reduced inference latency by 20% through hyperparameter tuning and architecture optimization. Collaborated with 5+ engineers to integrate and deploy ML models into production systems

CNNsOpenCVTensorFlowDeep Learning

AI Intern

Datics AILahore, PakistaninternshipJul 2024 – Sep 2024

Summary

Designed ML models for NLP applications using Python, Pandas, and Scikit-learn, achieving 85%+ classification accuracy. Built chatbot prototypes handling 500+ daily queries with automated response generation, reducing manual support load

Results

Achieved 85%+ classification accuracy

PythonPandasScikit-learnNLP

Skills

Technical

Python

TensorFlow

Git

GitHub

Scikit-learn

CSS3

Pandas

HTML5

Node.js

Express.js

REST APIs

MongoDB

JavaScript

SQL

Keras

OpenCV

NumPy

React

Next.js

MySQL

Tailwind CSS

Bootstrap

OOP

C++

Linux

Agile

Scrum

SOLID Principles

MVC

Flask

PHP

Docker

ONNX

Rapid prototyping

Supabase

Slot-locking

Voice models

KNN

Vercel

Projects

Guardian Sense

Architectpersonal

Architected an AI-driven emergency response system using smartphone sensors; TensorFlow classifier achieved 92% detection accuracy. Built cross-platform Flutter app and RESTful Node.js backend supporting 50+ concurrent users

Next.jsNode.jsFlutterMongoDBTensorFlowONNX

Check INN, Chill Out

Developerpersonal

Developed full-stack hotel management system with booking engine, role-based access control, and JWT-secured API endpoints

MERN StackJWTSupabase

Disease Prediction System

ML Developerpersonal

Trained 4 ML classifiers (Random Forest, SVM, KNN, Logistic Regression); achieved 88% accuracy on 5,000+ patient records using SMOTE

PythonScikit-learnPandasKNN

Spare Parts Management System

Developerpersonal

Built inventory system with real-time stock tracking; optimised indexed MySQL queries, reducing report generation time by 40%

PHPMySQLBootstrap

Education

Bachelor of Science

FAST National University of Computer and Emerging SciencesComputer ScienceSep 2022 – Jun 2026

Recognition

Certifications

Deep Learning Specialization (5 courses)

Coursera / DeepLearning.AI

Hire MuhammadReach out about a role, a contract or a conversation.For recruiters

Muhammad's twin is AI, it can make mistakes.

More people on Kanvis

Browse more people Create your own Kanvis