Abdullah Sajid

Full-stack Developer & Data Scientist | Bridging HCI, ML, and Product Design

Based in: Lahore, Punjab, PakistanMost recently: Freelance Web/App Developer, Self-EmployedExperience: 3 years

LUMS Computer Science student with a unique blend of full-stack development, data science, and brand design. From engineering credit scoring models at Techlogix to building end-to-end MVPs for startups and instructing government officers in ML, I specialize in creating data-driven, human-centered digital experiences. Currently focused on the intersection of HCI and AI-assisted development.

Experience

Freelance Web/App Developer

Self-EmployedPakistan, Australia, Canadawork2025

Summary

Built client websites and MVPs for startups and small businesses, owning projects end to end from product scoping and UI/UX design through frontend development, deployment, and iteration. Delivered marketing websites, e-commerce storefronts, product catalogs, checkout flows, and user-facing MVP workflows for construction, home services, bakery/retail, and technology clients. Worked directly with founders and business owners to translate rough requirements into launched products.

What I did

Conducted UI/UX audits and provided constructive design criticism to help founders refine their product vision and component architecture.

Results

Increased client acquisition by 10-15% for 'TheOvenEquation' (bakery) by integrating the website with Instagram marketing.
Developed a professional portfolio for 'Tameer Construction' (subsidiary of IZHAR Construction) to establish digital presence and brand authority.

UI/UX DesignFrontend DevelopmentE-commerceProduct Scoping

Course Assistant

LUMS X Federal Board of RevenueLahore, Punjab, PakistanworkOct 2025 – 2026

Summary

Instructed FBR (CSS/Government) officers under a LUMS-FBR MoU using Excel, Python, and ML tools for data analysis and visualization. Designed and graded quizzes, assignments, and tutorials across data science topics.

Results

Mentored FBR (CSS/Government) officers to resolve complex Python and ML misconceptions, facilitating critical 'aha' moments in data science applications for government workflows.

ExcelPythonML toolsData AnalysisData Visualization

Data Science Intern at Techlogix

TechlogixLahore, Punjab, PakistanworkJun 2025 – Aug 2025

Summary

Worked with large-scale financial datasets to build predictive models for credit scoring and fraud detection.

What I did

Engineered credit scoring and classification models using Scikit-Learn and Pandas.
Utilized diverse financial datasets including Income Survey Finance Analysis and Credit Card Transactions (Kaggle) to simulate real-world banking scenarios.
Optimized data retrieval via SQL queries and performed exploratory data analysis with Matplotlib and Seaborn.

Scikit-LearnPandasSQLMatplotlibSeabornMachine Learning

Skills

Technical

HTML/CSS

Python

Canva

VS Code

Git

Figma

OpenAI API

C/C++

JavaScript/TypeScript

React

Next.js

PyTorch

Scikit-Learn

Pandas

NumPy

Jupyter

Vercel

Adobe Creative Suite

Claude API

LLMs

Node.js

MATLAB

FastAPI

Matplotlib

Seaborn

Transformers

Flask

Supabase

SQL

React Native

Google Cloud Platform

Docker

Areas of expertise

Social Media Strategy

Brand Identity

Projects

R-SRD: Robust Skeleton Relation Distillation

Lead Researcheracademic

Proposed a 3-phase ViT distillation pipeline mining a sparse cross-domain consensus skeleton of token-token attention edges using a bilinear router probe and MMD-weighted agreement scoring for DG on PACS. Diagnosed catastrophic Sketch failure (49.1%) as a coverage deficiency; increasing skeleton density κ=0.2→0.3 recovered +13.3% accuracy, achieving the highest multi-seed mean of 84.0%.

PyTorchViTDomain Generalization

PerCVO: Text-to-Video Explainer Pipeline

Full Stack Developerpersonal

Created an automated pipeline that transforms text or PDF prompts into animated explainer videos. Integrated Claude API for content extraction and context mapping, using the Manim Python library for animations and AI-powered audio generation.

Claude APIManimPythonAI Audio

Carcinoembryonic Antigens Segmentation and Quantitative Analysis from Fluorescent Images using Principal Component Analysis and Adaptive K-means Clustering

Research Lead / Lead Developerpersonal

Developed a MATLAB pipeline using Superpixel analysis (SLIC), PCA, and Adaptive K-Means to detect and segment Carcinoembryonic Antigens from fluorescent scans. Achieved precise quantitative detection across a concentration range of 0-100ng/ml, with a T/C ratio of 1.0814 for high-accuracy biomarker identification. Published at IEEE ICEPECC 2023.

MATLABPCAK-Means ClusteringSLIC SuperpixelsImage Processing

Seekho Sikhao: Guard-railing AI for Early Education

Lead Developer / HCI Researcherpersonal

Developed an AI learning aid for children in low socio-economic status schools to prevent 'copy-paste' learning. Built a RAG pipeline using GPT-4o mini and Urdu LLMs, grounded in institutional policies and course content to ensure the AI provided guidance rather than direct answers.

OpenAI GPT-4o miniUrdu LLMsRAGPython

Media

PUBLICATION

IEEE Research Paper: CEA Segmentation & Quantitative Analysis

Full research paper on CEA segmentation using PCA and Adaptive K-Means, published at IEEE ICEPECC 2023.

Education

Bachelor of Science

Lahore University of Management SciencesComputer ScienceAug 2023 – Aug 2027

Ordinary & Advanced Levels

Lahore Grammar SchoolAug 2018 – Jun 2023

Recognition

Patents & publications

Carcinoembryonic Antigens Segmentation and Quantitative Analysis from Fluorescent Images using Principal Component Analysis and Adaptive K-means Clustering

patent2023

Published at IEEE ICEPECC 2023

With Muhammad Aqeel Aslam, Shahzadi Mahnoor, Muhammad Asif Munir, Saman Cheema, Khawaja Humble Hassan

Read it

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

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

More people on Kanvis

Browse more people Create your own Kanvis