Alyyan Awan

AI/ML Engineer with a strong foundation in production machine learning, deep learning, and MLOps, plus hands-on experience building agentic AI and LLM-based systems. Delivered production ML models (XGBoost, LSTM, Bi-LSTM, Physics-Informed Neural Networks) for EV battery analytics with a Ford-affiliated UK research team, achieving 91% prediction accuracy. Architected and rebuilt AuditX, a 6-agent multi-agent orchestration system using LangChain, LangGraph, and Retrieval-Augmented Generation (RAG), cutting inference latency from 8-10s to 3-5s via migration to vLLM.

Experienced across the full ML lifecycle — data preprocessing, feature engineering, model training, fine-tuning, experiment tracking, CI/CD, containerization (Docker), and cloud deployment (AWS) — with multiple end-to-end MLOps pipelines shipped to Hugging Face Spaces.

AI Engineer at iOPTIME Pvt Ltd (2025-11 – 2026-05)

• Built and deployed machine learning models (XGBoost, LSTM, Bi-LSTM, Physics-Informed Neural Networks) for EV battery analytics, trained on 1,200+ synthetic records derived from 64 original battery datasets, achieving 91% prediction accuracy in production.
• Developed FastAPI inference services for real-time EV battery predictions for a Ford-affiliated research team in the UK, including automated validation logic and diagnostic monitoring that eliminated manual analysis bottlenecks.
• Architected and rebuilt AuditX, a production 6-agent multi-agent orchestration system for AI-powered automated auditing, using LangChain and LangGraph.
• Implemented hybrid Retrieval-Augmented Generation (RAG) with page-level indexing to improve document retrieval accuracy, and migrated the inference backend from Ollama to vLLM — reducing response latency from 8-10s to 3-5s in production.
• Built end-to-end ML pipelines including data preprocessing, feature engineering, model training, and deployment; benchmarked multiple LLMs to optimize response quality and latency across the auditing platform.

AI & Machine Learning Engineer at CSERA (Private) Limited (2025-07 – 2025-09)

• Improved object detection accuracy by approximately 12% through systematic experimentation with YOLOv8 model weights, data augmentation strategies, and OpenCV preprocessing pipelines.
• Built and deployed real-time computer vision inference systems; integrated MLOps workflows using PyTorch, MLflow, and FastAPI for experiment tracking and model serving.

AI & Web Development Intern at Digital Bytes SMC Pvt. Ltd. (2024-06 – 2024-09)

• Built and deployed end-to-end machine learning pipelines for a retail client (Gala Groceries), from data preprocessing through model training to production deployment with MLOps integration.
• Developed and maintained web APIs using FastAPI and Flask to serve live machine learning model predictions.

Machine Learning Intern at Air University (2023-07 – 2023-09)

• Implemented anomaly detection algorithms on public network traffic datasets (CICIDS) for cybersecurity research, applying predictive analytics to classify threat patterns and benchmark detection accuracy across multiple ML models.

BS in Artificial Intelligence – Air University, Islamabad (2021-09 – 2025-06)