CV

Archon Tech / Senior Engineer

Nov 2025 – Present · Copenhagen, Denmark

Joined as a frontend specialist and grew into full-stack and platform ownership across a research-led product group building on-chain financial-risk infrastructure. Work spans four distinct areas, all shipped as production software rather than prototypes.

Data platform and risk modelling. Contributed across a Go/Python/TypeScript monorepo that indexes Ethereum and L2 state in real time, tracks DeFi lending and DEX positions, and computes required risk capital for institutional allocators. Built a second independent data path that reconciles the platform’s own on-chain figures against a counterparty’s self-reported balance sheet, replacing per-request live HTTP integrations with a properly indexed pipeline and eliminating a class of silent-degradation bugs. Rolled a Monte-Carlo risk model into production, including the capacity study — peak resident memory measured across nine markets — that set its pod sizing. Root-caused and fixed two production incidents others could not explain: a TimescaleDB hypertable recompressed without segment keys, defeating chunk pruning and pegging CPU on every query; and a connection pooler OOMing mid-transaction, leaving database connections desynced in a state standard health-checking could not detect.

Frontend architecture and a shared design system. Designed and built a TypeScript/React toolkit published as fifteen npm packages — a Panda CSS design system on headless primitives with a semantic token contract and AA-verified dark mode, a visx-based charting layer with cross-chart cursor synchronisation and large-series downsampling, a virtualised data grid, a schema-driven dashboard engine, and HTTP client and mock packages whose types are both derived from the same generated OpenAPI contract so mocks cannot drift from the client. Ran repeated bundle-size forensics on the result, measuring every module in isolation to find that one library barrel import cost 82 kB minified on its own, and restructuring the package into lazy tiers — so that a consuming application no longer had to hand-maintain its own chunk map to get its entry chunk down from 1,004 kB to 80 kB.

Infrastructure, observability and reliability. Built an AWS observability and paging stack end to end for a production conversational-AI service: CloudWatch alarms and metric filters derived from both request metrics and structured application logs, a Slack incident-paging pipeline on Lambda and Step Functions with re-notification and thread grouping, synthetic uptime canaries, and a self-hosted Grafana platform migrated off a managed offering once its sharing and version constraints blocked a business need. Tuned alarm thresholds against the observed log-normal latency distribution rather than naive percentiles. On the shared Kubernetes/GitOps platform, provisioned durable storage and workload identity for new tenants, sized autoscaling capacity with worked arithmetic, and governed alert noise by root-causing the underlying deploy behaviour. Repeatedly found alerting that looked correct but structurally could not fire — in one case quantifying that a staleness alert had been silently disarmed for weeks before rewriting it.

Language and systems research. Contributed to the core engine of a from-scratch logic and term-rewriting language — pattern matching, unification, non-deterministic rewriting under a typed discipline — including new language-level semantics for the boundary between data and executable code. Separately designed and built the compiler front end for a domain-specific financial-specification language: a parser compiling a restricted subset of a general-purpose language into symbolic-math expression graphs, a code generator emitting a typed importable client, and a graph-database-backed API and web UI for browsing the resulting knowledge graph with branch history, diffing and dependency visualisation. Built and operated an event-sourced simulation platform to stress-test that engine at production-like scale, including checkpointed boot, idempotent replay and a real-time analytics dashboard whose incremental fold cut per-event cost by roughly five orders of magnitude.

Engineering practice. A consistent method runs through all of it: measure before and after and publish the numbers; reproduce a bug before fixing it; sabotage-test a fix by reintroducing each historical defect to confirm exactly the intended test goes red; and treat tooling — build reproducibility, CI gates, dev environments, AI coding-agent configuration — as infrastructure held to the same correctness bar as the product.

Most work at Archon Tech is under NDA and is described here at the level of the problem rather than the product; the projects below are the parts that are public.

Edge Delta / Senior Front-end Engineer

Apr 2023 – Oct 2025 · Copenhagen, Denmark

Led the transition of the core product from Angular to React, and the adoption of a shared component library to go with it.

Extended Metrics Explorer — a Grafana-like query and charting surface — with support for multiple queries and formulas, and redesigned it around the new component system.

Built Service Map and Metrics Inventory for introspection into deployment architecture and data cardinality, and worked on the Monitors and Composite Monitors overhauls.

Delivered Kubernetes Metrics monitoring, shipped with a walkthrough demo.

Churney / Lead Front-end Developer

Oct 2022 – Feb 2023 · Copenhagen, Denmark

Worked directly with the founders to establish a web platform and dashboard giving clients self-service access to their models and insights.

Built embeddable interactive visualisations used for marketing and outreach.

DataRobot / Senior Front-end Developer

May 2019 – Sept 2022 · Copenhagen, Denmark

Joined the Geospatial Domain at an early stage to develop the inspection and visualisation capabilities that were integrated into the wider application.

Later joined the Batch Predictions team to build a job designer and scheduler for recurring prediction jobs, integrating with every supported data source and providing build-time validation and run-time error tracing.

Leapcraft / Data Visualisation & Full-stack Development

Jan 2015 – Apr 2019 · Copenhagen, Denmark

Built bespoke visualisations for consulting projects with large client data sets, generally using D3, Leaflet and Highcharts.

Grew into leading development and architecture of the cloud infrastructure for data ingestion, model execution and insight delivery on AWS — consuming real-time data from in-house sensor platforms and running ML models for interpretation and calibration.

Freelance / Creative Technologist

Jan 2013 – Dec 2014 · Copenhagen, Denmark

Consulted on several projects specialising in data visualisation and user experience, for clients including Leapcraft, KL7 and Kirt Thomsen — the first of which became a full-time role in 2015.

Also worked on Appeer, which received project funding through LaunchPad Denmark.

HITLabNZ / Research Assistant

Sept 2011 – Dec 2011 · Christchurch, New Zealand

Worked on augmented reality projects on the Android platform, alongside visual design for the lab’s output.

Purple Leaf Design / UX Designer & Developer

Nov 2009 – Dec 2009 · Guwahati, India

Worked on branding and identity alongside web development for the studio’s clients, with competitive research feeding both.

Zensar Technologies / Usability Engineering Intern

May 2009 – Jul 2009 · Pune, India

Interned with the Usability Engineering Group, producing wireframes, use cases and information architecture for enterprise software, and contributing to heuristic evaluations, personas and user interviews.