Available for work

Hello, I'm Maaz Haider

I build

Artificial Intelligence Engineer crafting intelligent systems, neural architectures, and ML pipelines that push the boundaries of what machines can do.

View My Work Let's Talk
0+ Models Deployed
0+ Years Experience
0+ Projects Shipped
Scroll
01 About Me
Maaz Haider
🧠 Deep Learning ⚡ PyTorch 🤖 LLMs 📊 MLOps
🏆
Top Performer
AI Competition 2024
🎓
MSc AI
Computer Science

Turning Data Into
Intelligence

I'm an AI Engineer passionate about designing systems that learn, adapt, and reason. From fine-tuning large language models to building end-to-end ML pipelines, I thrive at the intersection of cutting-edge research and production engineering.

My work spans computer vision, NLP, reinforcement learning, and generative AI — with a relentless focus on shipping models that actually perform in the real world.

🔬
Research-Driven
Papers → Production
🚀
Fast Iteration
Prototype → Deploy
🌐
Open Source
Contributor & Builder
Get In Touch
02 Skills

My Toolkit

Technologies and frameworks I work with daily

🤖 AI & Machine Learning
PyTorch / TensorFlow95%
LLM Fine-Tuning90%
Computer Vision88%
Reinforcement Learning80%
⚙️ Engineering & MLOps
Python92%
MLflow / Weights & Biases85%
Docker / Kubernetes82%
AWS / GCP / Azure78%
Hugging Face LangChain OpenAI API CUDA Scikit-learn FastAPI Pandas / NumPy Transformers Diffusers Triton Ray ONNX
03 Projects

Featured Work

A selection of AI systems I've built and shipped

LLM
01

Model Evaluation — LLM AS A JUDGE

A Docker-based framework that uses Large Language Models (LLMs) to automatically evaluate and score AI-generated responses. It enables consistent, reproducible benchmarking of model outputs using customizable evaluation criteria.

LLMLangchainKnowledge GraphFastAPIDocker
Computer Vision
02

AI Image Detector — Classify AI-Generated Images and Real Images

A deep learning model using EfficientNetB0 to classify images as real or AI-generated. Achieves ~93% accuracy, demonstrating effective detection of synthetic media.

EfficientNetB0TensorFlowKerasCUDA
Generative AI
03

RAG-GRAPH — Graph Retrieval-Augmented Generation

A graph-based Retrieval-Augmented Generation system that enhances LLM responses using knowledge graphs for structured and context-aware retrieval. Built with Python and LLMs, it applies graph traversal and semantic search strategies to improve reasoning and response accuracy.

DiffusersDDPMPyTorchW&B
Multimodal AI
04

SentiVision - Multimodal Sentiment Analyzer

Fuses RoBERTa text embeddings with CLIP ViT image patch features via cross-modal attention — letting each word attend to relevant image regions. Outperforms late-fusion baselines by +3% F1 on MVSA-Single. Ships with a FastAPI inference server, Gradio demo, and MLflow experiment tracking.

PyTorchCLIPRoBERTaFastAPIGradioMLflow
04 Contact

Let's Build Something Intelligent

Open to full-time roles, consulting, and exciting research collaborations. If you're working on something ambitious in AI — I'd love to hear about it.

✉️
Email
maazhaider0011@gmail.com
GitHub
github.com/maazhaider11
💼
LinkedIn
linkedin.com/in/maaz-haider-681546277/