Summary
Afaq A. is a highly skilled backend developer and data scientist with a strong academic foundation, currently pursuing a Bachelor of Science in Data Science from NUCES-FAST. He brings hands-on experience in Python, Django, AWS, AI, Git, Docker, Linux, Celery, Redis, SQL, React, JavaScript, VS Code, MySQL, and Tableau, alongside proficiency in advanced AI topics such as generative AI and NLP. At Uurnik Systems, Afaq has demonstrated success designing and maintaining RESTful APIs, building cloud automation solutions with Boto3, and integrating asynchronous task execution using Celery, significantly enhancing system performance and automation. His prior role as a lab demonstrator involved guiding students in SQL and Assembly, while his internship at Graana.com saw him leveraging Python, BeautifulSoup, Selenium, Tesseract, and NLP techniques for data extraction and translation tasks. Afaq's notable projects include ScholarWatch, where he developed an AI-powered Moodle plugin using LangChain and Gemini 2.5 LLM with React and PHP, as well as a research initiative combining Vision Transformers, Photoplethysmography, and cross-attention mechanisms for deepfake detection. Fluent in English and Urdu, with knowledge of German and Burushaski, Afaq excels in collaborative, agile environments and is adept at utilizing version control with Git and developing scalable, efficient backend systems.
Meet Afaq A.
Assessment Scores
Great on Implement LRU Cache Algorithm, Difficulty Medium
Great on Understanding Rate Limiting in APIs, Difficulty Medium
Great on Ensuring Real-time Data Accuracy in Mobile App, Difficulty Medium
Experience
Uurnik Systems
Backend Developer
Full-Time
Oct 2024 — Present
Collaborating in an Agile development environment to deliver software solutions.
Designing, developing and maintaining 20+ RESTful APIs and endpoints using Django for scalable backend services.
Developed an agentic RAG chatbot with a custom-built Model Context Protocol (MCP) server using Langgraph, automating ~70% of routine workflows and enabling dynamic task execution through conversational interfaces.
Built backend support for cloud automation using AWS (Customer Gateways, Virtual Gateways, Site-to-Site VPNs) via the Boto3 SDK, improving deployment reliability and security.
Implemented asynchronous task execution using Celery, offloading heavy computations and external API calls to background workers, reducing request latency significantly.
Developed and managed Celery Beat scheduled tasks for recurring jobs, improving system automation and reducing manual operational effort.
Implemented network automation for multi-vendor routers (Cisco, Juniper, Fortinet), enhancing efficiency and reducing manual configuration efforts time by ~60%.
Utilized Git for version control, contributing to 30+ commits/month in collaborative development.
Technologies:
Python
Django
AWS
AI
Git
Docker
Linux
Celery
Redis
NUCES-FAST
Lab Demonstrator
Full-Time
Feb 2024 — May 2024
Guided 60+ undergraduate students across 2 core courses (Databases and Computer Organization & Assembly Language).
Resolved 50+ student queries, improving student lab performance by ~20% through concept clarification and live debugging.
Technologies:
SQL
Assembly
Graana.com
Data Science Intern
Intern
Jun 2023 — Aug 2023
Scraped and processed 1000+ property listings using BeautifulSoup and Selenium for structured data extraction.
Automated OCR pipeline using Tesseract, increasing document data extraction speed by 40%.
Leveraged NLP methods to preprocess text and hypertune T5-small for English to Urdu translation.
Technologies:
Python
Projects
ScholarWatch
Collaborated in a team of 3 to build an AI-enhanced Moodle plugin with 7+ features; including gaze tracking, quiz generation, and weak-area analysis, providing smart engagement tools and personalized learning insights for remote education. Implemented RAG using LangChain's vectorStoreIndex with Gemini 2.5 LLM for generating personalized dynamic quizzes with 90%+ accuracy in topic alignment. Built an analytics dashboard for students and instructors, delivering personalized academic insights, weak-area detection, engagement metrics, and performance trends through real-time visualization of learning data. Utilized Git for version control to manage codebase collaboration and ensure structured development workflows.
Research: Multimodal Deepfake Detection using ViT, PPG & Cross-Attention
Co-authored a research paper, proposing a novel deepfake detection framework combining Vision Transformers (ViT) for facial analysis and Photoplethysmography (PPG) for physiological signal extraction. Integrated a cross-attention mechanism to align visual and pulse-based cues for improved detection accuracy. Achieved 82.6% accuracy on a subset of Celeb-DF-v2 (890 real, 890 fake videos), showcasing strong results with limited resources. Contributed to model design, dataset preparation, and co-writing of the research paper.
Education
Bachelor of Science (Data Science)
2021 - 2025
National University of Computer & Emerging Sciences (NUCES-FAST)
Islamabad, Pakistan
