Shahzaib Khan
AI/ML Engineer specializing in deep learning, computer vision, and generative AI with hands-on experience developing production-ready models. Proven expertise in transformer architectures, RAG systems, and LLM applications using PyTorch, TensorFlow, and LangChain.
Experience
AI Developer Intern
Revotic AI
Rawalpindi, Pakistan
Developed RAG-based chatbot using LangChain, Llama 3-8B, and FAISS achieving zero hallucinations through optimized retrieval pipeline and semantic search with 90%+ accuracy Engineered production NLP pipeline with Hugging Face embeddings and Groq API reducing response latency by 60% while handling 100+ concurrent sessions on cloud infrastructure Built scalable Streamlit UI supporting multi-document upload and real-time natural language querying with sub-2-second response time
Data Science Intern
ITSOLER Pvt.Ltd
Rawalpindi, Pakistan
Developed Stock Price Prediction models using machine learning and deep learning techniques, performing time-series preprocessing, feature engineering, and model evaluation to achieve stable predictive performance Built a Scotland Birth Prediction system leveraging historical demographic data, applying regression and forecasting methods for trend analysis and future estimation Implemented and compared multiple ML models with optimized training pipelines, including data normalization, hyperparameter tuning, and performance benchmarking, improving generalization and reducing overfitting
Education
Pakistan Institute of Engineering And Applied Sciences(PIEAS)
Bachelor of Science
Relevant Coursework: Deep Learning, Machine Learning, Computer Vision, NLP, Neural Networks, Data Structures, Algorithms
Skills
Projects
Deepfake Video Detection System
PyTorch, Vision Transformers, OpenCV
Architected end-to-end detection system combining ViT, ConvNeXt, and Swin Transformer with temporal optical flow features achieving 92% accuracy and 85% reduction in false positives Implemented custom attention mechanisms and Fourier-based feature extraction processing 10,000+ video frames with robust preprocessing pipeline using MTCNN for face detection
Bookverse AI Recommendation System
Python, Collaborative Filtering, Content-Based
Built hybrid recommendation engine combining collaborative and content-based filtering achieving 87% recommendation precision for 5000+ book catalog Implemented matrix factorization and cosine similarity algorithms with optimized data preprocessing pipeline
Computer Vision Classification Models
TensorFlow, CNNs, Transfer Learning
Trained ensemble CNNs using ResNet50, InceptionV3, and EfficientNet achieving 94% top-1 accuracy with transfer learning reducing training time by 70%
Certifications
AI Developer Internship
Revotic AI
Data Science Internship
ITSOLER Pvt.Ltd