summary
Recent Master's graduate in Data Analytics (University of Niagara Falls Canada, October 2026) with an M.Sc. in Physics and a strong grounding in statistics, mathematics and modelling. I build end-to-end analytics solutions in Python, SQL, Power BI and Excel: star-schema data warehouses, automated data-quality checks, KPI dashboards, demand forecasts and optimisation models. Currently a remote Data Engineer at Shipd (DataCurve), creating training data for AI models after being recruited on the strength of my Kaggle results (top 10 to 15% finishes in global competitions). Holder of five professional certificates in data analytics and business intelligence from Google, SAS and Microsoft, and an experienced writer who explains data clearly to non-technical audiences.
experience
Data Engineer (remote)
Feb 2026 – PresentShipd by DataCurve, San Francisco, CA, USA
- Recruited directly by the company after a top-15% Kaggle competition finish, to build high-quality training data for large language models and AI coding agents (Project Eris).
- Design complex programming and data challenges, with verified reference solutions, that expose reasoning gaps in advanced AI agents.
- Write, test and refine machine-learning code used to train and fine-tune models, meeting strict standards for correctness and quality.
Data Annotator (remote)
Jun 2025 – Sep 2026TELUS Digital, Canada
- Labelled and quality-rated multimodal AI training data (text, image and video) on the Multimango and Parimango platforms.
- Performed entity tagging on social-media content, video interleaving, and image and video quality rating against detailed guidelines, keeping accuracy consistent at volume.
Content Writer (full-time, remote)
Apr 2019 – Apr 2024BrightMedia International, Calicut, India
- Researched and wrote news articles, investigative reports, science-communication pieces and reviews, plus marketing copy, online ads and social-media content, to deadline for five years.
- Turned technical and scientific topics into clear, accurate writing for a general audience, the same skill I now use to present data insights to stakeholders.
data and AI projects
Python, SQL, Excel, PuLP
- Built an end-to-end BI pipeline on a synthetic dataset of five hotels (about 640,000 bookings over four years): star-schema warehouse, ETL, ten automated data-quality checks and unit tests.
- Designed a Revenue Command Centre dashboard covering occupancy, ADR, RevPAR, booking pace against last year, cancellations, channel net rates and a back-tested demand forecast.
- Developing revenue and cost analyses (group displacement, rate recommendations, overbooking, energy costs) and a staff scheduler that turns forecast workload into shifts with mixed-integer optimisation, checks Ontario labour rules and exports to Excel.
Python, PyTorch, XGBoost, CatBoost, LightGBM
- Mallorn Challenge (top 15% of about 800 teams): detected rare astronomical events in noisy time-series data with under 1% positive cases, using physics-informed features and an XGBoost and CatBoost ensemble tuned with Optuna.
- Name That LLM (436th of 4,346 teams, top 10%, ongoing): fine-tuned Qwen2.5-7B and GLM-4-9B with LoRA on a 38,935-passage training set and built leak-free grouped cross-validation (0.663 balanced accuracy).
- NeurIPS 2026 Smart Buildings: designed a physics-informed HVAC control policy that beats the best public notebook (0.01244 vs 0.01853, lower is better), and found and reported a bug in the organisers' simulator.
- Also: Squares in Squares (top 13%), Enveda CASMI 2026 molecule identification, CAFA protein-function prediction (deep learning with extreme class imbalance), and Google MedGemma Impact Challenge.
PyTorch, JavaScript, Leaflet.js, Gemini API
- Developed a decentralised drone-swarm simulation using a Deep Q-Network to route drones to wildfire choke points, building on my earlier multi-agent reinforcement learning model of ERCOT (Texas grid) energy data.
- Built a live web dashboard with satellite mapping and a Gemini-powered chatbot that explains the model's decisions in plain English.