I help businesses and teams look more professional, save time, and work smarter with bespoke websites, dashboards, automation, and internal tools. From 850,000+ records processed in daily operational pipelines to premium client websites, I build solutions that are reliable, measurable, and built for real-world use.
A quick snapshot of the proof, services, and business value behind the work featured on this site.
I'm Dorrian Sidwell — a data engineer and AI systems developer with 4+ years of production delivery inside a large public-sector organisation. I build systems that take messy, manual, high-stakes processes and turn them into reliable pipelines, dashboards, and operational tools people can actually trust.
That work has improved reporting, compliance, and operational forecasting by reducing manual effort, improving data quality, and giving teams clearer visibility into performance. My safeguarding compliance project also identified 1,499 incorrect mappings, reshaped how the organisation measures compliance, and earned a Merit distinction in my apprenticeship capstone.
Alongside core data engineering, I also build analytics dashboards, AI-powered support products, white-label internal platforms, and selected premium client websites. That breadth helps when a project needs someone who can understand the data layer, build the product, and ship the final experience without losing sight of business value.
The core tools I use to deliver websites, dashboards, data platforms, automation, and internal systems that work in practice — not just in demos.
A short technical sample for recruiters and technical reviewers who want to see implementation detail from live operational tooling.
import pandas as pd from rapidfuzz import process, fuzz from pathlib import Path # 16,000+ positions mapped and validated df_matrix = pd.read_excel( mapping_file, sheet_name="Active PNs", usecols=list(range(9, 18)) ) # Fuzzy match against validated list for col in mapping_cols: for idx, val in df[col].items(): match = process.extractOne( val, valid_courses, scorer=fuzz.ratio, score_cutoff=85 ) if match: df.at[idx, col] = match[0]
import psycopg2 # Stage-swap: 850k+ records via COPY conn = psycopg2.connect( dbname="ops_pipeline", host="localhost" ) with conn.cursor() as cur: cur.execute(f""" CREATE TABLE {stage} (LIKE {table} INCLUDING DEFAULTS) """) with open(csv_path) as f: cur.copy_expert( f"COPY {stage} FROM STDIN CSV HEADER", f ) # Atomic swap to production cur.execute(f"ALTER TABLE {table} RENAME TO _old") cur.execute(f"ALTER TABLE {stage} RENAME TO {table}") cur.execute("DROP TABLE IF EXISTS _old") conn.commit()
Examples drawn from large-scale operational delivery featured in this portfolio, showing how the underlying systems perform and why the outputs can be trusted.
A mix of live client work, production systems, and custom platform builds that show the kinds of problems I can help solve.
These are the projects that best show how I help businesses and teams improve online presence, reduce friction, and build more capable systems.
"Bespoke client website with self-service CMS, SEO setup, and premium brand presentation"
A bespoke website for a real client that improved brand presentation, simplified content updates, and gave the business a clean, maintainable platform for enquiries and growth. Built to feel premium, perform well, and hand over confidently to a non-technical owner.
"850,000+ records processed in daily operational pipelines · 8,255.5 staff hours of unnecessary training removed"
Four years of production data engineering inside a large healthcare organisation — automating reporting, rebuilding fragile data flows, and delivering ETL pipelines, PostgreSQL models, and Power BI dashboards that support real operational decisions across the organisation.
"Custom AI support platform with local-first deployment options and role-based workflows"
A multi-tenant AI helpdesk platform designed to reduce repetitive support workload without forcing organisations to send sensitive knowledge externally. Built with a 7-stage RAG pipeline, local-first Ollama support, role-based controls, and full white-label branding for commercial reuse.
"AI-assisted marine data platform for mapped observations, validation, and research use"
A citizen science platform designed to turn diver observations into structured marine biodiversity data. Combines AI-assisted species ID, mapped dive-site discovery, community verification, and research-ready export — with a strong focus on data quality rather than vague concept hype.
These projects show range, curiosity, and systems thinking, but they are intentionally separated from the primary professional case studies above.
"Trolls have stolen your mead! 19 scenes · 1,288 sprites · 17 custom extensions"
A Viking-themed 2.5D side-scroller built in GDevelop with custom Blender assets. Players battle through Norse realms (Midgard, Jotunheim, Alfheim, Helheim) to reclaim stolen mead. Features tutorial progression, mobile-first joystick controls, realm-based level selection, and a full custom AI-assisted 3D asset pipeline using Blender Python scripts and ComfyUI.
"ML-driven trading with XGBoost, LSTM, ARIMA & Prophet models"
An algorithmic trading platform with 4 ML models (XGBoost, LSTM, ARIMA, Prophet), 30+ technical indicator feature engineering, risk management (VaR 95%, Kelly criterion position sizing), async OHLCV data loading via CCXT, comprehensive backtesting engine, and performance metrics (Sharpe, Sortino, Calmar ratios, MFE/MAE analysis).
"Real-time odds arbitrage with AI-powered opportunity ranking"
A matched betting platform with real-time Betfair Exchange API integration, qualifying/SNR/SR bet calculators with lay drift recalculation, AI-powered opportunity ranking (heuristic + optional GPT re-ranking), fuzzy event matching across bookmakers, and automated Telegram notifications for profitable opportunities.
A clearer offer structure for businesses and teams that need a better online presence, less manual work, and stronger operational visibility.
Clear scope, decisive execution, and production-minded delivery.
We define the real problem, business goals, user needs, constraints, and success metrics before delivery starts.
Information architecture, technical approach, data structure, and UX direction are agreed up front so the build stays deliberate.
I build in focused milestones with visible progress, regular check-ins, and practical feedback loops rather than disappearing into a black box.
Performance, QA, responsive behaviour, deployment, and launch support are handled end-to-end so the final result is ready for real users.
Handover, documentation, iteration, and optional retained support keep the product maintainable and improving after launch.
Need a website, dashboard, or internal tool? Let's talk about what's slowing your business down and what would make the biggest difference. I'm open to freelance projects, contract work, and selected long-term opportunities.
Based in
Liverpool, United Kingdom
Available for remote work worldwide
Thanks for getting in touch. I'll get back to you within 1–2 business days.
Something went wrong — please try again or email me directly at soaddorrian@gmail.com.