ANAS.
ANASELRAHMANI
CS Student & AI Researcherturning models into products, and ideas into code.
Qatar University · Doha, Qatar · Expected 2027
Scroll
AI RESEARCHER ·DEEP LEARNING ·MULTIMODAL LLMs ·FULL-STACK DEVELOPER ·AWS CERTIFIED ·SAMSUNG ·HBKU ·QATAR UNIVERSITY ·AI RESEARCHER ·DEEP LEARNING ·MULTIMODAL LLMs ·FULL-STACK DEVELOPER ·AWS CERTIFIED ·SAMSUNG ·HBKU ·QATAR UNIVERSITY ·AI RESEARCHER ·DEEP LEARNING ·MULTIMODAL LLMs ·FULL-STACK DEVELOPER ·AWS CERTIFIED ·SAMSUNG ·HBKU ·QATAR UNIVERSITY ·AI RESEARCHER ·DEEP LEARNING ·MULTIMODAL LLMs ·FULL-STACK DEVELOPER ·AWS CERTIFIED ·SAMSUNG ·HBKU ·QATAR UNIVERSITY ·
AI Assistant
Ask meanything.

This assistant knows my work, skills, and background. Ask about a project, my experience, or what I'm looking for next.

Start a conversation, ask me about Anas's work, skills, or background.

About
Computer scientist,based in Doha,building where researchmeets the real world.

Who I am

I'm Anas Elrahmani, a Computer Science student at Qatar University, graduating in 2027. I focus on AI and machine learning, not just training models, but understanding what it actually takes to make them work on real problems. The part that interests me most is where research has to become something that actually ships, where getting the engineering right matters just as much as getting the math right.

Experience

At Hamad Bin Khalifa University, I fine tuned and benchmarked multimodal vision language models like DeepSeek VL and Qwen2 VL, achieving 15% higher accuracy and 30% faster inference. At Samsung Innovation Campus, I built an LSTM based fake news classifier that reached 98.75% accuracy across 40,000+ articles. I'm also AWS AI Practitioner certified, with hands on experience deploying generative AI solutions on AWS.

Philosophy

A model that works in a notebook is not a product. I care about the gap between the two: inference speed, real world reliability, and whether what you built actually helps someone. If you're working on something where the research and the product are equally hard to get right, I'd like to hear about it.

0
Classifier accuracy for 40k+ articles
0
Model accuracy in HBKU research
0
Faster inference on VL models
AWS
AI Practitioner certified
Selected Work
SelectedWork
04 Projects
01
SAMSUNG
Fake News Detector
Designed and trained an LSTM based fake news classifier achieving 98.75% accuracy. Preprocessed a corpus of 40,000+ articles using text encoding and tokenization techniques, then iteratively fine tuned model hyperparameters to improve generalization.
LSTMDeep LearningPython
01
SAMSUNG
Fake News Detector
01
PROJECT IDENTITY
SAMSUNG
KEY METRICS
Model Accuracy
98.75%
Corpus Size
40k+
LSTMDeep LearningPython
Fake News Detector
Designed and trained an LSTM based fake news classifier achieving 98.75% accuracy. Preprocessed a corpus of 40,000+ articles using text encoding and tokenization techniques, then iteratively fine tuned model hyperparameters to improve generalization.
02
HBKU
Multimodal LLM Research
Fine tuned and evaluated multimodal LLMs including DeepSeek VL and Qwen2 VL on large scale vision language benchmarks. Optimized training pipelines and preprocessing workflows, achieving 15% improvement in model accuracy and 30% faster inference speed.
DeepSeek VLQwen2 VLFine Tuning
02
HBKU
Multimodal LLM Research
02
PROJECT IDENTITY
HBKU
KEY METRICS
Model Accuracy
+15%
Inference Speed
30% faster
DeepSeek VLQwen2 VLFine Tuning
Multimodal LLM Research
Fine tuned and evaluated multimodal LLMs including DeepSeek VL and Qwen2 VL on large scale vision language benchmarks. Optimized training pipelines and preprocessing workflows, achieving 15% improvement in model accuracy and 30% faster inference speed.
03
QATAR UNIVERSITY
University Management App
Designed and implemented a full stack web application for managing students, instructors, and courses. Phase 1 built core features using HTML, CSS, and JavaScript with JSON based storage. Phase 2 migrated to a relational database with Prisma and Next.js, adding REST APIs and a statistics dashboard.
Next.jsPrismaFull Stack
03
QATAR UNIVERSITY
University Management App
03
PROJECT IDENTITY
QATAR UNIVERSITY
KEY METRICS
Architecture
Full Stack
Database
Relational
Next.jsPrismaFull Stack
University Management App
Designed and implemented a full stack web application for managing students, instructors, and courses. Phase 1 built core features using HTML, CSS, and JavaScript with JSON based storage. Phase 2 migrated to a relational database with Prisma and Next.js, adding REST APIs and a statistics dashboard.
04
PERSONAL
Portfolio 2026
This portfolio. A scroll driven experience built from scratch with Next.js, GSAP, and Lenis, featuring pinned card animations, smooth scroll transitions, and a custom cursor system.
ReactGSAPFrontend
04
PERSONAL
Portfolio 2026
04
PROJECT IDENTITY
PERSONAL
KEY METRICS
Animations
GSAP
Scroll
Lenis
ReactGSAPFrontend
Portfolio 2026
This portfolio. A scroll driven experience built from scratch with Next.js, GSAP, and Lenis, featuring pinned card animations, smooth scroll transitions, and a custom cursor system.
Capabilities
Skills &Expertise
AI & Machine Learning
  • PyTorch
  • TensorFlow
  • HuggingFace
  • Large Language Models
  • Fine Tuning
  • Vision Language Models
  • Computer Vision
  • MLOps
Web Development
  • React
  • Next.js
  • Node.js
  • TypeScript
  • Prisma ORM
  • REST APIs
Languages
  • Python
  • JavaScript
  • Java
  • SQL
  • MATLAB
  • HTML / CSS
Cloud & DevOps
  • AWS (AI Practitioner Certified)
  • Amazon Bedrock
  • Git
  • Docker
  • Jupyter Notebook
  • VS Code
Course Certificates
01
AWS AI Practitioner Challenge
AWS AI Practitioner Challenge
02
Neural Networks and Deep Learning
Neural Networks and Deep Learning
03
Samsung Innovation Campus
Samsung Innovation Campus
CONTACT
Get in touch
Let's workon something.
LocationDoha, Qatar