I’ve been building for the web since 2005, starting with UI design before moving into JavaScript full-time. Node.js
drew me into full-stack development. Tutoring maths and physics beforehand honed my mentoring instincts.
My career grew into platform architecture, team leadership, and XP practices - I love inviting feedback early,
whether from my IDE or stakeholders. Currently exploring how AI agents can accelerate development, building
multi-agent systems with automated quality gates. I remain drawn to engineering leadership: shaping teams,
refining processes, and ensuring quality at scale.
Currently using: TypeScript, React, Claude Code, MCP, AWS, Kubernetes. Previously: D3.js, Rails, Clojure, OpenShift, GCP.
Best code is no code. Small, incremental, transparent, continuously verified and integrated changes come second.
Marek Stasikowski
Experience
Kotai London
Lead Engineer
- present
Building AI-powered developer tooling and MCP-based automation systems, focusing on
practical applications that deliver measurable productivity gains.
comms-mcp: A unified communications assistant connecting email, iMessages, and WhatsApp
via Model Context Protocol. Processes 280 emails/week with AI summaries, reducing triage
time by 70%[1]. Used to resolve a small claims dispute:
instant message retrieval, evidence organisation, and automated court correspondence
recovered £1,080.
hybrid-memory: A semantic memory system combining pgvector embeddings with BM25
full-text search via Reciprocal Rank Fusion, enabling long-running agentic sessions to share
insights across projects.
Staff engineer at an omnichannel retail platform, focusing on system resilience,
performance optimisation, and developer productivity.
Unblocked $6M ARR product add-ons feature by pragmatically isolating one component from a
stalled "big rewrite", using characterisation tests to safely lift it to production - cutting
delivery time from months to 2 weeks.
Designed circuit breaker patterns for both synchronous (HTTP) and asynchronous (message queue)
flows to prevent incidents caused by third-party systems, each costing up to 5 hours of
on-call time plus week-long follow-ups.
Led BFCM load testing and optimisation: tuned Kubernetes HPA, scaled AWS databases, optimised
SQL queries - achieving target of 44 orders/second (over 2x previous year's average).
Prototyped NewStore MCP - connectors to GitLab, AWS, OpenSearch, Prometheus, RabbitMQ, and
Kubernetes - reducing incident investigation from up to a day to 20 minutes[2]
by unifying data scattered across microservices, Slack, Confluence, and observability tools.
Presented at weekly knowledge sharing sessions on legacy codebase strategies, testing approaches,
and Monte Carlo-based delivery forecasting.
Platform engineer for an 800-engineer monorepo, focused on CI performance and onboarding efficiency.
Championed Blacksmith migration to halve 30+ minute merge queues and reduce CI costs 4x -
design approved, taken over by security review and procurement.
Quantified onboarding friction: 30 days to get "hello world" to production. First automation
iteration estimated to save 5 days per new project.
Coached teams in mutation testing using Stryker to expose gaps and overlaps in test coverage,
providing this as a new platform capability.
Joined as consultant to streamline a 100k+ SLOC React/Redux onboarding application;
permanently hired as Principal Engineer after one year.
Shaped architecture of multi-dimensional onboarding app, clearing significant tech debt
while championing TDD, BDD (Cypress, Cucumber), and robust CI/CD practices.
Led migration from legacy Java monolith to serverless AWS using facade pattern, gradually
replacing legacy flows with Lambda functions and Step Functions.
Designed system integration with SingPass and KYC providers like Signicat.
Implemented Options trading extensions in the user account section, contributing to
multi-million ARR feature rollout within 3 months.
Zero missed FCA regulatory deadlines - critical for a financial services platform.
Frequent mentorship unblocking team members across frontend and backend work.
Built customer-facing platform configuration and admin tooling; promoted to Team Lead December 2021.
Established XP feedback loops: fast tests, acceptance tests, pair programming, daily deploys
with QA/PO involvement, and fortnightly stakeholder product reviews.
Reduced cycle time variability from 50% to near zero - tickets consistently completed in 2 days.
Replaced finger-in-air estimates with Monte Carlo projections based on historical data,
removing time pressure from engineers.
Reduced time spent on ceremonies: standups under 10 minutes using ActionableAgile aging WIP view,
focusing only on outliers and current technical decisions.
Coached team in TDD; organised practice sessions across London engineering community.
Engineer for an espionage museum's interactive experiences and visitor profile system in New York.
Extended games and online profiles with new ways of browsing takeaway videos and
exploring visitor data.
Worked with real-time data streaming from physical devices (RFID wristbands, iPads)
throughout the venue via AWS Kinesis.
Technologies used:AWS:EC2KinesisS3RDSAPI GatewayRuby on RailsNode.jsReactRedux
Equal Experts
Software Consultant
-
JavaScript consultant at major UK retailer and Swedish fintech.
John Lewis & Partners
-
Built foundation for customer support app targeting reduction of £800k/year
live agent costs; mentored JL Partners in frontend development.
Cleared CSS theming tech debt to enable rebranding in My Account. Later built
Auth0 integration replacing legacy system.
Technologies used:ReactReduxAuth0
Klarna
-
Joined Authentication & Personalisation team to take ownership of services inherited
from another team, reverse engineering and providing out-of-hours support.
Technologies used:Ruby on RailsNode.js
uSwitch
Senior Developer
-
Rebuilt legacy Rails energy switching journey using React/Node with Strangler Fig pattern.
Improved conversion rate from 2-3% to nearly 4% through modernised UX and A/B tested
incremental migration.
Redirected traffic step-by-step to new stack, eventually reducing Rails app to pure
data service.
Technologies used:Node.jsClojureRuby on RailsGraphQLReactRedux
notonthehighstreet.com
Front End Engineer
-
R&D role pioneering server-side React rendering for checkout pages.
Established universal React approach that became industry standard; set technical
strategy for the business.
Shared learnings at meetups (NOTHS, Pivotal, University of London) and ReactJS Day Verona.
Technologies used:Node.jsReactRedux
Earlier Experience
McKinsey & Company/JavaScript Consultant
-
Worked directly with international clients, fitting solutions to existing infrastructures. Ran a one-stop shop for front end/JavaScript prototypes, both mobile and desktop.
Node.js · AngularJS
Cigna Insurance Services/JavaScript Developer
-
Built travel and life insurance policy configurators in a team of four AngularJS developers.
AngularJS
Pearson PLC/JavaScript Developer
-
Developed a Node.js API to query Elasticsearch and built UIs for a big data analysis platform in the Data Analytics and Visualization team.
Node.js · Backbone.js · Elasticsearch
SiteSell Inc./JavaScript Developer
-
Worked remotely with a Canadian team on a website building solution, developing rich MVC components including file managers with drag-and-drop functionality.
Vanilla JS · Custom MVC
Roche Pharmaceuticals/Front End Developer
-
At the corporation's main software house, designed information architecture and built UI mockups and front ends for internal applications.
K2 Internet/Web Developer
-
First full-time role at the largest interactive media agency in Poland, building websites for various clients.
References
↑ 70% triage reduction: Based on 280 emails/week processed with AI summaries. Previously
required manual review of each email (est. 2 min average); now only flagged items need attention
(est. 30% of total). Assumes consistent email volume and complexity.
↑ 20 minute investigation time: Measured across 5 incident investigations using the MCP
prototype vs. previous manual process. Investigation typically required switching between GitLab
(code, pipelines, MRs), AWS Console (CloudWatch, RDS, Lambda), OpenSearch (logs), Prometheus/Grafana
(metrics, dashboards), RabbitMQ Management (queue depths, message rates), Kubernetes (kubectl, pod logs),
Slack (incident channels, historical context), Confluence (architecture docs, runbooks), and service-specific
internal documentation to manually confirm hypotheses about root causes or understand feature behaviour (e.g. order routing logic).