Muhammad Subhan Shahid
Aspiring AI / Machine Learning Engineer
AI-focused undergrad skilled in LLMs, MLOps, and Computer Vision. Experienced in building end-to-end AI systems and seeking AI/ML internships.
About
AI-focused undergraduate with hands-on experience in computer vision, machine learning, NLP, and model deployment. Skilled in convolutional neural networks, transformer architectures, reinforcement learning, and automating ML pipelines. Proven ability to design end-to-end AI systems and lead complex university projects. Actively seeking AI/ML internships.
Skills
Languages: English
Education
Bachelors in Artificial Intelligence
FAST National University of Computer and Emerging Sciences
Intermediate in Computer Science (ICS)
Unique College, Lahore
Matriculation in Computer Science
Unique School, Lahore
Experience
Private Academic Tutor · Self-Employed
2021 – PresentManaged a concurrent roster of 6 students over a 5-year period, demonstrating strong time management while simultaneously completing a rigorous BS Artificial Intelligence degree. Developed strong communication and mentoring skills by translating complex academic concepts into easily understandable lessons. Fostered a track record of reliability, adaptability, and independent problem-solving.
Projects
Agentic Dialogue Response Generation
Transitioned standard Seq2Seq text generation into an Agentic AI framework. Fine-tuned a quantized Llama-3 (8B) model using LoRA/PEFT to enhance contextual dialogue and integrated it within a LangChain/LangGraph architecture.
View project →Real-Time Market Movement Prediction System
Architected a comprehensive MLOps pipeline for tracking and automating model training workflows. Utilized MLflow, DVC, Docker, Makefiles, and GitHub Actions to deploy on AWS.
View project →Conversational AI with RAG and Lightweight Agentic Tool Integration
Engineered a Retrieval-Augmented Generation (RAG) pipeline integrating a FAISS vector database to dynamically retrieve context from a custom knowledge base.
View project →Computer Vision for Classifying Pakistani Politicians
Developed an end-to-end computer vision application to accurately classify images of political figures. Deployed using FastAPI and Streamlit.
View project →Fake Face (Deepfake) Detection System
Developed a computer vision model utilizing Convolutional Neural Networks (CNNs) to accurately distinguish between real and AI-generated human faces.
View project →Fake News Detection System
Built a classification pipeline using transformer models (BERT, RoBERTa, DistilBERT) to process and detect fabricated news.
View project →Text Summarization using Encoder-Decoder Models (BART, T5)
Implemented abstractive text summarization using Transformer-based encoder-decoder architectures on the arXiv dataset.
View project →Phishing Email Detection & Deployment
Trained and evaluated Logistic Regression, FNN, RNN, and LSTM models for identifying phishing emails, deployed through a Gradio/Streamlit interface.
View project →