Eidetic

Independent AI product and systems engineer · London

Robin Miklinski, drawn in stipple and dissolving into a field of data points.

Robin Miklinski

I build AI products, then try to break them.

Fifteen years of experience delivering and validating systems and building engineering teams across global platforms, retail, defence and security.

Discuss a project → See selected achievements ↓

Previously

  • Reward Gateway Employee engagement platform
  • ASOS Global online fashion retailer
  • Perkbox Employee benefits and wellbeing platform
  • Huddle Secure document collaboration platform
  • BAE Systems Detica Cybersecurity and intelligence consultancy
  • Steria Defence UK Defence technology consultancy
  • HP Computing and printing technology
  • Fujitsu Digital services and IT systems

Selected results 01—03

Evidencing AI Quality Reward Gateway · Employment history
I developed a large automated suite of prompt injection, tenant isolation and personal data tests across web, iOS and Android.
Growing an engineering team whilst rebuilding its delivery infrastructure Reward Gateway · Employment history
The function grew from 4 to 15 engineers, supporting more than 200 people in product and engineering.
Rebuilding an inherited planning applications scoring system Your Right 2 Light · Client engagement
I re-architected a large planning application database, restoring council datasets and rebuilding the scoring system.
01 / Achievements

Selected achievements

Reward Gateway Employment history · 2019–2026

Reward Gateway: engineering systems at organisational scale

  • AI product evaluation. Evaluation combined reference answers, repeated runs and regression gates for the company’s first customer-facing AI product.
  • Incident analysis. A Python pipeline analysed hundreds of production incidents over ninety days and supported a £227,000 annual business case for preventable failure.
  • Delivery infrastructure. I built and optimised the test pyramid running in CI, including automated functional and non-functional tests (pytest, Cypress, JMeter load tests).
Employment history Read the Reward Gateway case study →
Your Right 2 Light Client engagement · 2025–2026

Your Right 2 Light: rebuilding and improving an inherited production application

  • Rebuilt the scoring system and integrated planning data in an inherited Dart, Flutter and Firebase application.
  • Delivered application notes and batch reassignment features, alongside tooling for production data corrections.
  • Restored a failed council dataset, re-scored more than 24,000 planning applications and onboarded the external development team.
Client engagement
02 / About

Background and experience

Engineering. Head of Quality Engineering at Reward Gateway, with hands-on work on AI evaluation and release readiness.

  • Experience spans defence and intelligence data systems, retail at peak scale, collaboration software and global SaaS used by millions.
  • Work is primarily in Python, JavaScript, TypeScript and C#. Qualifications include a BSc (Hons) in Internet Engineering from the University of Exeter.

Tools in practice

Selected technologies across current product work and earlier platform delivery.

AI systems and agents
AWS Bedrock, OpenSearch, Ollama, Claude Code, Codex, MCP
Product engineering
Python, JavaScript, TypeScript, C#, SQL, React, Dart, Flutter, Firebase
Testing and assurance
pytest, Playwright, Cypress, Selenium, SpecFlow, Detox, Maestro, JMeter, Locust, k6, OWASP ZAP, axe, BrowserStack
Cloud and delivery
AWS, Azure, Cloudflare, Amazon SQS, Docker, Kubernetes, Terraform, GitHub Actions, GoCD, TeamCity, LaunchDarkly
Data and observability
SQLite, Redshift, Datadog, New Relic, Tableau, Heap
Workflow integrations
Jira, Confluence, Zapier

Speaking. A 2024 talk at BrowserStack in London, Observable Test Automation at Scale, covered scaling test automation and GenAI-assisted pull-request review.

Outside engineering. Interests include audio engineering and electronic music production. Previous work includes a residency at EGG London and hosting a DJ show on AAJA Radio.

03 / Selected code

Tools from my music workflow

eidetic-sample-tools

A Python toolkit combining local AI, audio analysis and hardware integration for music production.

  • Local AI classification. Audio–text models classify samples, with cached embeddings and a listening interface for reviewing uncertain results. The classifier is experimental.
  • Audio search and curation. Search combines musical role, character and acoustic similarity with audition playlists and curated collections.
  • Hardware integration. Exports support Octatrack, Digitakt and TR-8S, with format and capacity checks. Content hashes verify sample identity; original audio is preserved.
Source →

audio-service

An audio-processing service for metadata cleaning, BPM detection and FLAC-to-AIFF conversion, with source retention enabled by default.

Source →
04 / Contact

Tell me about your project.