Skip to main content
Portrait of Jonathan Collins

Jonathan Collins

AI and full-stack software consultant for energy trading and financial markets

Birmingham, UK

I bring artificial intelligence to the energy and finance markets, building software that creates genuine business value. I integrate AI into trading, risk and reporting workflows, turn market data into evaluated models, and engineer full-stack cloud solutions with Azure, JavaScript, C# and Python.

Technical consultancy

Expertise

Hands-on technical consultancy that brings AI into energy trading, financial markets and finance, from agentic development workflows and AI integration to data analysis, cloud architecture, and frontend and backend engineering. Each card sets out what I offer, the technologies involved and the experience behind it.

AI development

AI

AI-assisted software development, and AI built into everyday business workflows in energy trading, financial markets and finance.

  • Agentic development workflows
  • Custom AI instructions and skills for coding agents
  • AI and workflow integration across energy trading, financial markets and finance

Technologies

  • Microsoft Foundry
  • Copilot
  • Claude

Experience

months

1 year 6 months

Experience: 18 months (1 year 6 months)

AI data analysis

AI

Turning raw and market data into analysis, statistical models and trained models for AI applications.

  • Data acquisition and capture for AI applications
  • Statistical modelling
  • Time-series analysis of market data
  • Model learning, including training and evaluation
  • Data visualisation and reporting

Technologies

  • Python
  • Pandas

Experience

months

1 year

Experience: 12 months (1 year)

Technical architecture

Technical

Designing and documenting cloud and container-based solutions.

  • Cloud solution architecture
  • Containerised applications with Docker
  • Container orchestration with Kubernetes
  • Technical documentation in PowerDesigner

Technologies

  • Azure
  • Docker
  • Kubernetes
  • PowerDesigner

Experience

months

5 years

Experience: 60 months (5 years)

Frontend development

Technical

Web applications, from UI design through to single-page and isomorphic implementations.

  • Single-page applications (SPAs)
  • Isomorphic applications, rendered on the server and in the browser
  • UI design in Figma

Technologies

  • React
  • Angular
  • Vue
  • Figma

Experience

months

15 years

Experience: 180 months (15 years)

Backend development

Technical

The APIs, messaging and databases behind applications.

  • API development in C#, JavaScript and Python
  • Messaging with Azure Service Bus and Kafka
  • Document and columnar data with Azure Cosmos DB and Databricks
  • Relational databases with SQL Server and PostgreSQL

Technologies

  • C#
  • JavaScript
  • Python
  • Azure Service Bus
  • Kafka
  • Azure Cosmos DB
  • Databricks
  • SQL Server
  • PostgreSQL

Experience

months

15 years

Experience: 180 months (15 years)

Discuss a consultancy engagement

Projects

Showcases

Selected projects for Shell Power, Sembcorp, Schroders and Digital Realty, showing where these skills have been put to work. Explore all the showcases to filter by domain and technology.

  • Client: Shell Power

    Energy trading

    CRATOS

    Renewable energy

    Duration

    months

    Duration: 18 months (1 year 6 months)

    Front office aggregation and trade execution application for renewable energy.

    • Trade aggregation and execution
    • React
    • Zustand
    • SignalR
    • LaunchDarkly
    • Vite
    • Cypress
    • +13 more technologies, listed in the details
  • Client: Digital Realty

    Technical

    Global Portal

    Global provisioning platform

    Duration

    months

    Duration: 12 months (1 year)

    Global user management and authentication portal.

    • API development
    • Messaging systems
    • React
    • Redux
    • Next.js
    • Web Components
    • Vue
    • TypeScript
    • +6 more technologies, listed in the details
  • Client: Sembcorp

    Energy trading

    Auto Trade

    Energy trading, multiple exchanges

    Duration

    months

    Duration: 24 months (2 years)

    Sembcorp's first automated trading application, using layered algorithms to execute trades across multiple exchanges.

    • Trade aggregation and execution
    • Single-page applications
    • API development
    • TypeScript
    • Angular
    • single-spa
    • Jasmine
    • C#
    • ASP.NET Core
    • +8 more technologies, listed in the details
  • Client: Schroders

    Financial markets

    Solvency II

    Asset management, regulatory reporting

    Duration

    months

    Duration: 12 months (1 year)

    Migrated legacy database-driven applications that distribute Solvency II regulatory reports to single-page web applications.

    • Regulatory reporting
    • Single-page applications
    • API development
    • ES6
    • React
    • Redux
    • webpack
    • Babel
    • Material UI
    • +7 more technologies, listed in the details
View all showcases

Insights

AI Insights

Where AI can make the greatest difference, and what it can realistically save. Each insight sets out what I can offer, the skills behind it and, where available, the published research or official documentation it draws on.

Cost efficiency from AI-assisted development

AI

In randomised field trials with 4,867 developers at Microsoft, Accenture and a Fortune 100 company, developers with an AI coding assistant completed 26% more tasks (Cui et al., 2025). For the same team cost, that means more delivered work, or the same work in less time.

What I can offer

  • Identify where AI assistance gives your team the best return
  • Set up Copilot and Claude with instructions and skills so the gains hold across your codebase
  • Estimate the value for your team with the savings estimator

Source: Cui et al. (2025) (opens in a new tab)

Bigger gains for less experienced developers

AI

The same trials found that less experienced developers adopted the AI assistant more, and gained more from it (Cui et al., 2025). Instructions and skills can pass senior developers' knowledge to the whole team.

What I can offer

  • Capture your team's standards and domain rules in instructions and skills
  • Give newer developers AI support that follows them

Source: Cui et al. (2025) (opens in a new tab)

AI-automated software delivery

AI

AI coding agents can plan, build, test and review changes in agentic workflows. People stay in control and approve every change.

What I can offer

  • Assess where agentic workflows fit your delivery process
  • Set up agent workflows with test and review gates
  • Help your team work alongside AI agents

Learn more: GitHub Docs: Copilot on GitHub (opens in a new tab), Claude Code overview (opens in a new tab)

Making AI assistants follow your standards

AI

Repository instructions and reusable skills give Copilot and Claude your architecture, conventions and domain rules. Their output becomes consistent and easier to review.

What I can offer

  • Write instruction files for your codebase
  • Build a library of reusable skills for common tasks
  • Keep them current as the code changes

Learn more: GitHub Docs: repository custom instructions (opens in a new tab), Claude Agent Skills (opens in a new tab)

AI integration with Microsoft Foundry

AI

Microsoft Foundry brings AI models into your own applications and processes. Integration works best when it is designed into the application, not bolted on.

What I can offer

  • Design and build Foundry-based integrations in your existing applications
  • Connect AI features to your services, data and workflows
  • Deliver them in JavaScript, C# or Python

Learn more: Microsoft Learn: What is Microsoft Foundry? (opens in a new tab)

Getting the most from Copilot and Claude

AI

Copilot and Claude work best when each is used deliberately, with shared instructions and skills. That gives a team consistent results whichever assistant is used.

What I can offer

  • Advise where each assistant fits your team's workflow
  • Set up both with shared instructions and skills

Learn more: GitHub Copilot documentation (opens in a new tab), Claude Code overview (opens in a new tab)

AI savings

Estimate what AI could save your team

An illustrative estimate grounded in published research, not a promise. Adjust the inputs to reflect your team and see how much time AI-assisted development could free up; nothing you enter leaves your browser.

Example value; change it to match your team.

50%

Example value; change it to match your team.

26%

Default from Cui et al. (2025): 26% more completed tasks. Read the study (opens in a new tab)

Example value; change it to match your team.

Estimated yearly saving

About developers' worth of time

Value of time freed
£0
AI tool cost
£0

Discuss your estimate

Illustrative estimate, not a guarantee or a quote. It uses a published research average; real results depend on your team, codebase, tasks and adoption, and the estimate leaves out set-up, training and review time.

Source: Cui et al. (2025), The Effects of Generative AI on High-Skilled Work: Evidence from Three Field Experiments with Software Developers (opens in a new tab): 26.08% more completed tasks (standard error 10.3%) across 4,867 developers.

Counter-evidence: METR (2025) (opens in a new tab): AI tools made experienced open-source developers 19% slower on their own repositories in early 2025. METR 2026 update (opens in a new tab): Suggests a speed-up since then, but METR describes that newer data as unreliable.

The calculation runs entirely in your browser. Inputs are never stored, sent anywhere or tracked.

Contact

Get in touch with Jonathan Collins

Whether you are taking your first steps with AI or looking to get more from it, I would be glad to talk it through. Get in touch by email, phone or LinkedIn.

Available for technical consultancy enquiries.