AI Engineer & Architect with experience building and deploying production-ready AI systems using Generative AI, LLMs, and Retrieval-Augmented Generation (RAG). Expert in AI System Design, Multi-Agent Orchestration, and Enterprise-Scale Infrastructure. Skilled in developing robust, secure, and cost-efficient AI pipelines with a focus on reliability, scalability, and corporate-grade deployment (RBAC, Multi-Tenancy). Passionate about bridging research and production.

• AI Application & GenAI Development: LLM-based Applications, AI Chatbots & Assistants, Retrieval-Augmented Generation (RAG), Prompt Engineering & Prompt Optimization, Multimodal AI (Text + Vision), Model Inference, Integration & Evaluation.
• Data & MLOps Foundations: Data Pipelines & Feature Engineering, Model Training, Fine-Tuning & Inference, Model Evaluation (F1, ROC-AUC, RMSE), Experiment Tracking & Reproducibility, Handling Imbalanced & Large-Scale Data.
• Machine Learning & Deep Learning: Supervised & Unsupervised Learning, Deep Neural Networks (DNN, CNN, RNN, LSTM, GRU), Transformer Models, NLP & Computer Vision, Time Series Forecasting, Transfer Learning & Fine-Tuning.
• Deployment, Cloud & DevOps: Docker, Containerized AI Applications, Microsoft Azure App Service, Azure AKS, OpenShift, VPS, IIS, Vercel, Netlify, GitHub, GitLab CI/CD, Google Play Console, Microsoft App Center.
• Mathematics & Core Foundations: Linear Algebra, Probability & Statistics, Optimization Techniques, Gradient Descent, Activation & Loss Functions.

• Built and deployed end-to-end AI applications, including data pipelines, model training, fine-tuning, inference, and user-facing integration.
• Developed LLM-powered chatbots and RAG systems, applying prompt engineering to improve accuracy, contextual understanding, and response quality.
• Implemented multimodal AI solutions combining text and image inputs using deep learning and transformer-based models.
• Exposed trained ML/DL models through RESTful APIs using Flask and FastAPI, and built interactive interfaces with Streamlit and Dash.
• Deployed and scaled AI applications using Docker, Azure App Service, Azure AKS, OpenShift, Vercel, Netlify, VPS, and IIS.
• Managed version control and automated deployments using GitHub and GitLab CI/CD, with application releases via Google Play Console and Microsoft App Center.

Working on a deep learning research project using the CheXpert chest X-ray dataset. Focused on developing and validating models for multi-label chest pathology detection, applying advanced computer vision and deep learning techniques to improve medical image analysis and support clinical decision-making.

Applied data analysis and computational techniques to process experimental datasets, strengthening skills in statistical modeling and data-driven research. Used Python-based analysis and mathematical modeling to extract patterns and insights, supporting research workflows relevant to machine learning applications.

Architected a multi-tenant AI platform for automated contract analysis and risk assessment. Features include Role-Based Access Control (RBAC), logical tenant isolation, rate-limiting, and structured audit logging for corporate-grade security. Tech Stack: FastAPI, PostgreSQL, Tesseract/ViT, Docker, Kubernetes Readiness.

A master curation of architectural blueprints for Advanced RAG, Multi-Agent Orchestration, Scalable Inference, and LLM Cost Optimization. Includes trade-off analysis for vector databases and production-ready K8s deployment patterns. Tech Stack: Mermaid.js, Technical Architecture, System Design.

Architected a production-grade observability platform for monitoring LLM systems. Tracks prompt versions, token/cost efficiency, real-time hallucination rates, and P95 latency with automated LLM-as-Judge evaluation. Tech Stack: FastAPI, PostgreSQL, Streamlit, Plotly, SQLAlchemy.

Developed a production-grade multimodal system integrating ViT/CLIP Vision Encoders with Mistral-7B LLMs. Features Grad-CAM explainability, cross-modal reasoning, and autonomous structured report generation for high-stakes domains. Tech Stack: PyTorch, CLIP, Mistral, OpenCV (Grad-CAM), FastAPI.

Built a complete, production-grade MLOps system with automated retraining, drift detection (KS tests), and model versioning via MLflow. Scalable infrastructure designed for high-availability predictive services. Tech Stack: XGBoost, MLflow, FastAPI, Docker, SciPy.

Built a full end-to-end LLM fine-tuning pipeline for domain-specific instruction tuning. Optimized for memory-efficient training on 24GB GPUs using QLoRA and NF4 quantization. Deployed as a high-performance inference server with latency benchmarking. Tech Stack: Python, HuggingFace (PEFT/TRL/BitsAndBytes), PyTorch, FastAPI, Docker.

Launched a production-grade RAG blueprint featuring Hybrid Search (BM25 + RRF), Cross-Encoder Reranking, and an automated observability dashboard for cost and faithfulness monitoring. Tech Stack: LangChain, FAISS, Sentence-Transformers, Streamlit, Groq/OpenAI.

Architected a multi-agent orchestration system where specialized agents (Planner, Researcher, Executor, Critic) collaborate autonomously to solve complex tasks with self-correction and memory integration. Tech Stack: LangGraph/Custom Orchestrator, CrewAI patterns, Pydantic, Redis.

Developed an intelligent medical chatbot capable of retrieving accurate, context-aware responses using Retrieval-Augmented Generation. Integrated Hugging Face embeddings with Pinecone for global-scale semantic search. Tech Stack: Python, LangChain, Pinecone, FastAPI, Docker.

Integrated interactive dashboard for stock trend prediction using sentiment analysis and ensemble deep learning models (ARIMA, LSTM, XGBoost). Tech Stack: Python, XGBoost, Scikit-learn, Streamlit.

• MS Data Science | Air University Islamabad | 2025 - Ongoing
• BS Mathematics | Namal University Mianwali | 2020 - 2024
• Deep Neural Network Bootcamp | GIK Institute Swabi, Pakistan | June 2024 - Aug 2024

1. View-Aware Design for Efficient Deep Learning in Chest X-Ray Interpretation: Proposed a deployment strategy for chest X-ray AI by stratifying CheXpert into frontal and lateral views.
2. A DFT study of structural, electronic, mechanical, phonon, thermodynamic, and H2 storage properties of lead-free perovskite hydride MgXH3(X=Cr, Fe, Mn).
3. Hydrogen Storage Capacity of Lead-Free Perovskite NaMTH3 (MT=Sc, Ti, V): A DFT Study.