Birmingham, England, United Kingdom
Experienced Data Engineer / ML Engineer with 25+ years of expertise in data warehousing, analytics, and machine learning. Proven track record of optimizing large-scale Teradata environments, focusing on performance tuning, query optimization, data migration, and ecosystem management. Extensive hands-on experience with Teradata tools and technologies, including Data Mover, Query Grid, Clearscape. Skilled in implementing Teradata backup and recovery solutions, managing Unix/Linux-based systems, and developing automation through shell scripting. Strong background in data analytics and machine learning — building and deploying models for regression, classification, and deep learning using Python, scikit-learn, TensorFlow, and Keras. Proficient in developing AI-driven applications using Streamlit, Django, and integrating Retrieval-Augmented Generation (RAG) and Large Language Models (LLMs) via Ollama and Hugging Face. Driven by curiosity and precision, I specialize in transforming complex datasets into scalable, high-performance solutions that enable data-driven decision-making and innovation.
· Developed a deep learning-based computer vision solution (MonReader) to detect page flipping from image sequences, enabling assistive reading workflows for visually impaired users. Built and evaluated multiple CNN architectures including ResNet18, EfficientNet, and a custom TinyVGG using PyTorch. Implemented sequence-level prediction logic using frame aggregation (threshold and ratio-based detection) to improve robustness. Applied transfer learning, fine-tuning, and early stopping to optimise model performance and reduce overfitting. Created evaluation pipelines and visualisation tools to compare model performance across different datasets and configurations. Achieved **99.84% accuracy** and **0.9983 macro F1-score** on a page classification task using transfer learning with ResNet-18, demonstrating high-performance computer vision modeling.
- Developed a deep learning-based computer vision solution (MonReader) to detect page flipping from image sequences, enabling assistive reading workflows for visually impaired users. Built and evaluated multiple CNN architectures including ResNet18, EfficientNet, and a custom TinyVGG using PyTorch. Implemented sequence-level prediction logic using frame aggregation (threshold and ratio-based detection) to improve robustness. Applied transfer learning, fine-tuning, and early stopping to optimise model performance and reduce overfitting. Achieved 99.84% accuracy and 0.9983 macro F1-score on a page classification task using transfer learning with ResNet-18, demonstrating high-perf CV modeling. - Built an end-to-end AI-driven talent search and ranking system using modern ML and MLOps practices. Implemented a FastAPI backend with semantic search powered by FAISS and sentence transformers, enabling high-performance vector similarity search. Developed an interactive Gradio GUI for user querying and result visualization. Containerized the application with Docker and deployed it to AWS ECS (Fargate) using a Docker Hub image, supporting scalable REST endpoints and production-grade serving. Led exploratory data analysis and model experimentation in Jupyter notebooks, including embedding workflows and ranking logic making use of FastAPI, FAISS, sentence-transformers, Gradio, Docker, AWS ECS, Python, vector search, NLP. - Built a scalable machine learning pipeline that achieved ~96% accuracy for term-deposit prediction by performing data preprocessing, feature engineering, cross-validation, and hyperparameter tuning using Python, Pandas, NumPy, Scikit-learn, and CatBoost. - Engineered a machine learning solution to predict customer satisfaction for a logistics startup, improving prediction accuracy to ~73% by performing data preprocessing, EDA, feature engineering, and model optimisation using scikit-learn, and delivering insights that enhanced customer retention strategies.
Extensive handson experience in Installation/maintenance/troubleshooting of High end Teradata MPP platforms including hardware, software (OS & Database), Data Migration, System expansion, integration etc. Also, providing remote support to other team members & customers country wide.
Installation, upgrade, HW/SW maintenance of NCR S series servers running MPRAS unix, AT&T unix, SCO etc.
In house LAN/WAN support and end user PC/Application support