I build structure from the chaos of fast-paced products.

Executive Summary

Led the end-to-end vision, strategy, and execution for a client collaboration portal. Built an enterprise-grade AI analytics layer with Role-Based Access Control (RBAC) to unlock benchmark insights while protecting client data privacy.

The Business Challenge

Client reporting was manual and fragmented, while fulfilment workflows created operational delays. The goal was to introduce real-time analytics and client-facing AI capabilities without exposing sensitive cross-client performance data or opening systems to prompt injection risks.

API & Core Systems Integration

• Integrated Microsoft Dynamics 365 APIs to automate order fulfilment and material collection.
• Embedded bi-directional Power BI APIs to surface live campaign performance data directly within the client portal.

Enterprise AI & Guardrail Engineering

• Tiered RAG Architecture: architected Retrieval-Augmented Generation pipelines using strict RBAC rules, ensuring AI responses only retrieved context matched to the client's specific subscription tier.
• Input & Output Guardrails: implemented real-time sanitisation layers to block prompt injections, jailbreaks, and out-of-scope queries before reaching the model, plus post-processing checks to prevent hallucinations and unverified data leaks.
• Automated Evaluation Pipelines: established continuous testing frameworks using synthetic evaluation datasets to benchmark model performance on answer relevance, context recall, and groundedness before deployment.

Business Impact & Commercial ROI

• Saved an average of 32 hours per product delivery in fulfilment team efficiency.
• Successfully deployed a zero-trust AI assistant that safely delivered automated performance benchmarking to thousands of users across multiple client tiers.

Executive Overview

Pivoted a complex, high-effort sales automation proposal into a lean, value-driven lead capture product. Delivered a fully functional, personalised solution in 8 weeks, generating £168,000 in revenue during its first month while establishing higher-margin upsell pathways for the sales team.

The Business Challenge

The sales leadership team proposed a fully automated website chatbot. The initial spec aimed to pull user login details, automatically generate bespoke product recommendations, create binding contracts, and spin up fulfilment projects inside the Microsoft Dynamics 365 ERP system automatically.

While theoretically impressive, this scope carried major risks:

• High Technical Friction: integrating automated contract generation and immediate ERP project creation required months of custom engineering work.
• Flawed User Assumption: the scope assumed buyers knew exactly what product tier they needed without human guidance. In reality, clients were often unclear on product capabilities and required consultation to select the right package.
• Missed Commercial Opportunity: fully automating the purchase process removed the opportunity for sales reps to consult, build trust, and sell higher-value packages.

Strategic Product Leadership & Execution

Rather than building an over-engineered system, I challenged the core assumptions and proposed an iterative, hypothesis-driven approach:

• Pivoted Scope to Value-First Input: replaced the automated chatbot with a streamlined, multi-step lead form. Before showing product recommendations, the form asked visitors to identify their primary business challenge (such as audience growth or customer retention) — validating user intent while gathering actionable data for sales reps.
• Integrated Lead Capture Architecture: connected the frontend form directly to Microsoft Dynamics 365 via APIs. Instead of auto-generating static contracts, the system created qualified sales opportunities with full context attached.
• Empowered Human-Centric Sales: designed the workflow to route leads instantly to account executives, who received pre-qualified user context to initiate warm outreach, conduct discovery calls, and uncover upsell potential before issuing tailored agreements.

Commercial Impact & Lessons

• Rapid Speed to Market: reduced engineering build time from a projected multi-quarter initiative to an 8-week release.
• Immediate Financial Returns: generated £168,000 in direct revenue within the first 30 days of launch.
• Improved Sales Efficiency: gave sales teams higher-intent, pre-qualified leads rather than cold contacts, significantly boosting discovery call conversion rates.

Executive Overview

Transformed an in-house experimentation and personalisation initiative by architecting an autonomous agentic framework using Opal AI solutions. Eliminated operational bottlenecks by automating historical data analysis, test configuration, execution, and reporting — reducing test setup time from 8 hours to mere minutes.

The Business Challenge

While the foundational goal was to establish a data-driven experimentation and personalisation platform from scratch, the initial implementation faced major operational hurdles:

• High Operational Drag: setting up a single experiment manually required up to 8 hours of effort across data, product, and engineering teams.
• Analysis & Metric Bottlenecks: determining the right target metrics and analysing historical performance data manually slowed down initial test creation.
• Scalability Limits: the heavy manual workload restricted the volume of experiments the team could run, limiting the speed of product personalisation iterations.

Strategic Product Leadership & Execution

To solve these inefficiencies, I led the development and integration of an AI agent ecosystem designed to handle the end-to-end testing pipeline:

• Opal AI Solution Integration: embedded Opal AI models into the core platform to process historical analytics and automate test creation.
• Autonomous Agent Architecture: designed specialised AI agents to perform dedicated tasks:
  • Data Analysis Agent: examined historical performance data to identify high-impact conversion opportunities.
  • Setup & Metric Agent: defined primary metrics, configured test parameters, and completed technical setups automatically.
  • Monitoring & Reporting Agent: executed tests, monitored statistical validity, and dispatched real-time status updates and summaries to team channels.
• Governance & Reliability: implemented validation guardrails to ensure AI-generated experiment parameters adhered to platform rules and statistical standards before launching live.

Measured Impact & Commercial ROI

• Massive Velocity Boost: cut test setup time from 8 hours down to minutes, enabling the team to run significantly more experiments concurrently.
• Hands-Free Reporting: automated progress notifications and performance updates, removing manual reporting tasks for product managers.
• Scalable Personalisation: unlocked rapid, continuous iteration for personalisation workflows across core digital properties.