
Insight-driven Data Analyst with 7+ years of experience delivering analytical products and data insight services to support insight-led decision making, planning, and service improvement. Experienced in working within multi-disciplinary, user-centred, agile environments, partnering closely with stakeholders to translate complex data into clear, actionable insight for non-technical audiences. Strong background in data management, data visualisation, statistical analysis, data quality, and governance, with proven ability to manage competing priorities and deliver high-quality analytical outputs to agreed timescales. Ambitious about improving data literacy, adopting pragmatic approaches, and supporting evidence-based transformation through digital service initiatives. Innovative thinker with a commitment to professional development and professionalism in all aspects of work.
Project 1: Speech-to-Text AI System for Smart Assistants, Technologies: Python, Flask, PyTorch, OpenAI Whisper, Wav2Vec2, Docker, AWS, JavaScript, Developed an AI-powered speech-to-text system using Whisper and Wav2Vec2 models for real-time transcription., Created an interactive web app for users to record voice commands and receive instant transcriptions., Deployed the system on AWS with Docker for scalability and cloud integration., Integrated voice command recognition for smart assistants, enabling hands-free interaction., Model Comparison: Whisper achieved better accuracy (WER: 10.2%) but slower speed (15s/60s audio) compared to Wav2Vec2 (WER: 12.5%, Speed:10s/60s), making Whisper ideal for accuracy and Wav2Vec2 for low-latency tasks., Project 2: Stroud Valleys Dipper Project with University of Gloucestershire, Developed a WordPress website with an interactive mapping system to visualize bird sightings using Leaflet.js for map rendering and Google Maps API for geospatial data integration., Designed and implemented an ETL pipeline using Pandas and SQL for efficient extraction, transformation, and loading of large datasets, resulting in faster data processing. Used QGIS for spatial data analysis and visualization to identify patterns in bird behavior., Key quantifiable measure: Improved data processing efficiency by 40% and reduced data update time by 30%., Project 3: Performing analysis on Crime in UK from year 2014-2018. (Using R), Performed predictive analysis using 5+ models, including Linear Regression, K-means clustering, Hierarchical clustering, k-NN classification, Random Forest classification, and ANOVA, achieving 85% prediction accuracy., Analyzed and visualized 1 million crime records, uncovering key crime trends across regions and years.