Muhammad Haseeb

Software Engineer | C++ & Low-Level Systems | Full-Stack AI Developer

Overview

Software Engineering graduate from COMSATS with a focus on low-level systems and AI-driven applications. Developed a multi-threaded DPI engine in C++ capable of real-time domain blocking and an AI-powered skin condition detection platform with 80% accuracy. Experienced in Java/Spring Boot through HPE simulations and proficient in MERN stack development.

Lahore, Pakistan

Project Portfolio

PERSONAL

Present

HPE Software Engineering Job Simulation - Employee Data API

Developed a RESTful API to handle employee data using Java and Spring Boot as part of the HPE virtual job simulation, focusing on backend service architecture and endpoint implementation.

JavaSpring Boot

PERSONAL

Present

FYP-FacialDermaAI

Full Stack Developer

AI-powered mobile & web platform that uses machine learning to detect facial skin conditions such as acne, eczema, rosacea, and hyperpigmentation from a photo upload, and recommend personalized treatments. Designed and implemented REST APIs for authentication, image upload, and prediction results using Node.js and Express following an MVC architecture, applying OOP principles while collaborating within a 3-member team. Integrated ML inference APIs built with Python and FastAPI, including an image-processing pipeline, achieving 80% classification accuracy across five skin-condition types. Conducted regular code reviews and technical troubleshooting to maintain model accuracy and improve overall code quality. Developed a cross-platform mobile application using React Native, implementing secure user authentication and a streamlined workflow for capturing and uploading facial images. Integrated backend APIs for image submission and diagnosis retrieval, enabling 80% accurate predictions across 5 skin condition categories.

Node.jsExpress.jsPythonFastAPIReact NativeREST APIsMVCOOP

Code

PERSONAL

Present

Network Traffic Analyzer (DPI Engine)

Backend Engineer

Deep-packet-inspection engine that classifies and blocks live network traffic by application, using TLS SNI extraction — without decrypting any traffic. Built a C++ engine that parses raw Ethernet/IPv4/TCP/UDP headers and extracts TLS Server Name Indication (SNI) from ClientHello handshakes to classify encrypted HTTPS traffic by destination domain, without decrypting any data. Designed a multi-threaded pipeline (dispatcher + worker pool) using consistent hashing on each connection’s five-tuple to guarantee thread-safe, lock-free flow tracking; verified zero data races using ThreadSanitizer. Extended the engine to intercept and block live traffic in real time on Windows using WinDivert; debugged and resolved real-world evasion paths (QUIC/UDP fallback, multi-domain CDN footprints, stateless flow classification) and documented the remaining architectural limitation (HTTP/2 connection coalescing) rather than overstating coverage.

C++WinDivertCMakemulti-threadingThreadSanitizer

Code