Kyle Roberts

Crafting exceptional digital experiences with clean code and thoughtful design

React
TypeScript
Next.js
Node.js
Tailwind
Available for projects
Get To Know Me

About Me

Crafting digital experiences with passion, precision, and purpose

đź‘‹Hello, I'm Kyle Roberts

An AI Development Lead and full-stack builder I ship useful, secure AI into production fast: real workflows, measurable impact, and tight feedback loops. 8+ years at JD Sports turning complex ops problems into simple products teams actually use.

I design practical AI-driven products with Next.js/TypeScript, Supabase (pgvector), and GCP/Vercel—shipped fast, secure by default, and measured by outcomes (AHT, adoption, CSAT, ROI).

Day-to-day: LLM agent assist, email/task automation, and trustworthy RAG—with GDPR/DPIA baked in and real enablement (champions, playbooks, dashboards). Ship fast, measure, iterate. No theatre—just useful software.

Core Technologies

AI IntegrationReactNext.jsTypeScriptNode.jsGCPVercelSupabaseAzureUI/UX Design

AI Development Lead

JD Sports Fashion plc

Jan 2024 - Present

Owning end-to-end AI solution delivery for Customer Care & Operations. Shipped LLM-powered workflows that reduced AHT with measurable yearly capacity gains and strong agent adoption.

Multi-Minute AHT Reduction

Junior System Administrator

JD Sports Fashion plc

Jun 2023 - Jan 2024

Managed Contact Centre applications ensuring high availability and optimal performance. Collaborated with cross-functional teams to identify automation opportunities.

Contact Centre Systems Expert

Let's Connect

kyleroberts69@live.co.uk
kyleroberts.co.uk
Rossendale, United Kingdom

Achievements

Custom Built AI Copilot for Customer Care
Multi Minute AHT Reduction per contact
PDF AI Extraction + Teams Approval

Available for Hire

Open to New Opportunities
Open to Relocation
Remote & On-site Available
Tech Arsenal

Technology Stack

Technologies I use to bring ideas to life

Backend

Node.js (API Routes/Edge)

Expert
3 years exp

Supabase Edge Functions

Intermediate
2 years exp

Frontend

React

Expert
3 years exp

Next.js

Expert
3 years exp

TypeScript

Expert
3 years exp

JavaScript

Expert
3 years exp

Tailwind CSS

Expert
3 years exp

Database

PostgreSQL

Expert
3 years exp

Supabase

Expert
3 years exp

Firebase

Expert
2 years exp

Redis

Intermediate
2 years exp

Cloud & DevOps

AWS

Intermediate
2 years exp

GCP

Intermediate
2 years exp

Firebase

Expert
3 years exp

Docker

Intermediate
1 year exp

Git

Expert
5 years exp

AI & Integration

OpenAI

Expert
3 years exp

LangChain

Expert
1 year exp

Vercel AI SDK

Expert
1 year exp

Vector Database

Intermediate
1 year exp

Microsoft Services

Power Automate

Expert
7 years exp

Power Apps

Intermediate
2 years exp
Open for full-time

From idea to impact, fast

Outcome-driven builds with compliance, reliability, and speed

Professional Journey

Experience

My professional journey and the impact I've made across different domains

January 2024 — Present

AI Development Lead

JD Sports Fashion plc

Own end-to-end AI solution delivery for Customer Care & Operations, from discovery through prototype to production, aligning with InfoSec and Data teams.

Key Achievements

Shipped LLM-powered email/task generation and assisted workflows that reduced handling time with multi-minute AHT savings per contact
Built secure RAG pipelines on Postgres/pgvector with embedding/search strategies, caching, and prompt/tooling patterns
Defined comprehensive governance: data minimisation, PII handling, access controls (OIDC SSO, RLS), prompt logging/evals
Rolled out internal enablement (playbooks, office hours) and tracked usage/adoption with KPI dashboards
Drove measurable yearly capacity gains and significant agent adoption

Technologies & Skills

OpenAI APIAnthropicTypeScriptSupabasepgvectorGCP Cloud RunVercelTerraformDatadogGrafanaPower Automate
June 2023 — January 2024

Junior System Administrator

JD Sports Fashion plc

Responsible for day-to-day management of Contact Centre applications, ensuring high availability and optimal performance of all systems.

Key Achievements

Ensured high availability and optimal performance across all Contact Centre systems
Collaborated with cross-functional teams to identify automation and process optimization opportunities
Acted as go-to technical expert for IT-related issues within the contact centre
Provided timely and effective solutions to both hardware and software challenges
Contributed to improved customer service metrics through system optimization

Technologies & Skills

System AdministrationContact Centre SystemsProcess AutomationIT InfrastructurePerformance Optimization
December 2019 — June 2023

Customer Care Senior Adviser

JD Sports Fashion plc

Led continuous-improvement initiatives including reporting, coaching, SOPs, and KPI tracking (CSAT, AHT), while handling complex escalations end-to-end.

Key Achievements

Led continuous-improvement initiatives covering reporting, coaching, and SOP development
Tracked and optimized key performance indicators (CSAT, AHT)
Handled complex customer escalations end-to-end with high resolution rates
Partnered with managers to standardise best practices across the team
Ran observation and feedback loops that improved quality and consistency

Technologies & Skills

Customer ServiceKPI ManagementProcess ImprovementTeam CoachingQuality Assurance
March 2017 — December 2019

Customer Care Adviser

JD Sports Fashion plc

Delivered first-line customer support across voice and non-voice channels, resolving complaints at first escalation and supporting team projects.

Key Achievements

Delivered exceptional first-line customer support across multiple channels
Resolved complaints effectively at first escalation stage
Maintained high customer satisfaction scores consistently
Supported team projects and collaborative initiatives
Built strong foundation in customer service operations

Technologies & Skills

Customer SupportMulti-channel CommunicationComplaint ResolutionTeam Collaboration
8+
Years at JD Sports
4
Role Progressions
AI
Development Lead
Enterprise AI Solutions

Production Systems

Real-world AI implementations delivering measurable impact in Customer Care and Operations

Enterprise AI
LiveOngoing since Jan 2024

Agent Assist Copilot

AI-Powered Customer Care Automation

Enterprise-grade AI copilot that transforms customer care operations by automating repetitive email/task generation with LLM-powered drafts and contextual recommendations. Achieved multi-minute AHT savings per contact with measurable adoption across teams.

JD Sports Fashion plc
Ongoing since Jan 2024

Key Features

  • Prompt-engineered templates with dynamic tool-calling capabilities
  • Contextual grounding from live CRM and order management data
  • Automated draft generation with suggested next actions
  • PII redaction and data minimization at ingestion layer
  • OIDC SSO integration with row-level security (RLS)
  • Comprehensive prompt logging with automatic redaction
  • Continuous evaluation pipeline for quality assurance
  • Real-time usage dashboards and KPI tracking

Deliverables

  • Production AI copilot deployed across Customer Care teams
  • Secure RAG pipeline with pgvector embedding search
  • DPIA-compliant data handling and governance framework
  • Multi-provider LLM orchestration with model failover
  • Datadog monitoring and Grafana analytics dashboards
  • Internal enablement playbooks and training materials
  • Measurable AHT reduction per contact

Technology Stack

OpenAI APIAnthropic ClaudeTypeScriptPythonSupabasepgvectorGCP Cloud RunVercelPower AutomateDatadogGrafanaOIDC SSO
Enterprise AI
LiveOngoing since Jan 2024

Agent Knowledge Search (RAG)

Intelligent Document Retrieval System

Production RAG system that consolidates fragmented SOPs and FAQs into a unified, searchable knowledge base. Leverages hybrid retrieval strategies with pgvector embeddings to deliver faster, more accurate answers and reduce escalations.

JD Sports Fashion plc
Ongoing since Jan 2024

Key Features

  • Hybrid retrieval combining keyword and dense vector search
  • pgvector-powered semantic embeddings for accurate matching
  • Re-ranking algorithms to optimize result relevance
  • Strategic caching layer for frequently accessed content
  • Human feedback loop for continuous improvement
  • Access-scoped collections with team/role-based permissions
  • Content redaction for sensitive information
  • Versioned source tracking with full audit trails

Deliverables

  • High-performance RAG search engine in production
  • Supabase/Postgres infrastructure with pgvector extension
  • Custom embedding and re-ranking pipeline
  • Usage and accuracy analytics dashboards
  • Access control framework with RLS policies
  • Content versioning and audit system
  • Reduced escalation rates and reassignments
  • Improved answer accuracy metrics

Technology Stack

SupabasePostgreSQLpgvectorPythonTypeScriptOpenAI EmbeddingsRe-ranking ModelsNext.jsVercelDatadog
Internal Tools / Enterprise Ops
Live6 weeks

Contact Centre Wallboard

Multi-screen, governed content displays

Centralised wallboard platform that runs like an automatic slideshow across multiple screens. Supports text, images, iframes (to preserve existing KPI wallboards), and more—without disrupting current metrics. Includes content zoning for sensitive areas, full approval workflows for all changes, and real-time remote control via an admin console.

JD Sports Fashion plc
6 weeks

Key Features

  • Multi-screen playlists with timed rotations
  • Content types: text, image, iframe, announcements
  • Non-disruptive KPI embedding via iframes
  • Zoning rules to restrict sensitive content per location
  • Full approval workflow for add/edit/delete
  • Role-based access with RLS
  • Real-time sync and remote control (admin console)
  • Audit logging, versioning, and rollback
  • Scheduling & expiry windows for content

Deliverables

  • Production wallboard web app for large displays
  • Content model & playlist scheduler
  • Zoning & location-based policy engine
  • Approval workflow and reviewer UI
  • Admin console for remote control and live overrides
  • Supabase Realtime integration for instant updates
  • Audit trail, version history, and governance reports
  • Seamless embedding of existing KPI wallboards

Technology Stack

Next.jsTypeScriptSupabase (Postgres, RLS, Realtime)Vercel (Edge/Hosting)WebSockets/Event streamsTailwind CSS
Process Automation
Live3 months

Automated PDF Data Capture

Power Automate Workflow Automation

Intelligent PDF processing automation that eliminates manual data extraction from inbound documents. Integrates Power Automate, SharePoint, and Teams adaptive cards to save hours weekly while reducing keying errors and accelerating downstream processing.

JD Sports Fashion plc
3 months

Key Features

  • Automated PDF parsing and field extraction
  • Data normalization and validation logic
  • Excel/Database integration for structured storage
  • Teams adaptive card approval workflows
  • Optional Azure OCR for scanned documents
  • Webhook callbacks for real-time status updates
  • Role-based access controls for file handling
  • Comprehensive approval audit logs

Deliverables

  • Production Power Automate flow deployed enterprise-wide
  • SharePoint/OneDrive integration for document management
  • Excel and Graph API connectivity
  • Teams adaptive card approval interface
  • PII handling standards and compliance measures
  • Link-scope security controls
  • Hours saved weekly on manual data entry
  • Significant reduction in data entry errors

Technology Stack

Power AutomateSharePointOneDriveExcelMicrosoft Graph APITeams Adaptive CardsAzure OCRWebhook APIs
Enterprise AI / Internal Tools
Live2 months

VAT Invoice Issuance (Centralised)

From spreadsheets to a governed, scalable app

Replaced multiple unversioned Excel files with a single, governed application for VAT invoice issuance. Agents follow a guided flow with live validations, integrations, and versioned templates. Rules and values are centrally managed and updated in real time—improving consistency, scalability, and auditability.

JD Sports Fashion plc
2 months

Key Features

  • Guided issuance flow (not fully automated)
  • Centralised config: VAT rates, rules, templates
  • Versioned templates and change history
  • Live validations against order/payment data
  • Role-based permissions and approvals
  • Admin UI for real-time updates to rules/values
  • Integration-ready connectors (Order/Payments APIs)
  • Comprehensive audit trail and event logging

Deliverables

  • Single source of truth application (replacing Excel)
  • Order/Payments API integration layer
  • Versioned, VAT-compliant PDF template engine
  • Access control and approval workflow
  • Admin console for dynamic rules/values
  • Cloud Run/Vercel deployment with monitoring
  • Audit and retention aligned to DPIA
  • Measured reduction in finance support tickets

Technology Stack

TypeScriptNode.jsNext.jsSupabase (Postgres, RLS, pgvector optional)Order Management APIPayment APIPDF generation librariesCloud RunVercelDatadog
Productivity Tool
Live1 month

Smart Bookmarks for Agents

Contextual Navigation & Knowledge Management

Intelligent bookmark system that eliminates context-switching by providing agents with tagged, searchable, team-specific presets surfaced contextually within the copilot interface. Accelerates resolution times and improves onboarding consistency.

JD Sports Fashion plc
1 month

Key Features

  • In-app bookmark management with smart tagging
  • Full-text search across bookmarks and metadata
  • Per-team preset configurations
  • Contextual surfacing within copilot workflows
  • Row-level security by team and role
  • Change history tracking for preset modifications
  • Least-privilege access controls
  • Lightweight usage telemetry for optimization

Deliverables

  • Production bookmark system integrated with copilot
  • Next.js/TypeScript frontend application
  • Supabase backend with RLS policies
  • Edge caching for performance optimization
  • Team and role-based access framework
  • Usage analytics and telemetry dashboard
  • Reduced handling time on lookup tasks
  • Improved agent onboarding speed

Technology Stack

Next.jsTypeScriptSupabaseRow-Level Security (RLS)Edge CachingVercelReactTailwind CSS
Collaborations & side projects welcome

Got an idea? Let’s prototype it together

Open to short engagements, proofs of concept, and advisory work—especially AI agents, RAG, and internal tools

Contact

Let's Work Together

Ready to bring your ideas to life? I'm always excited to work on interesting projects and collaborate with amazing people. Let's create something extraordinary together.

Or reach out directly: