SELECTED WORK

Ideas are easy. Making them real is the interesting part.

I'm Bjarni Kristján Leifsson — I work across Software Engineering, DevOps, Integration and Technical Project Management. Some of these systems I designed and built from the ground up. On others my role was integration and rollout: taking a system into a live environment and adapting it to local technical and operational requirements until it held up in real use. Both are on this page, and each card says which.

  • Product Development
  • Cloud & Infrastructure
  • DevOps & Delivery
  • Data & Integrations
  • Monitoring & Observability
  • Geospatial & Environmental
  • Aviation & Safety Systems
  • Technical Project Management
  • User-Centered Solutions
  • Research & Innovation

THE WHOLE SYSTEM — 7 LAYERS

Most engineers live on one or two of these. Most outages come from the ones they don’t.

Good technology solves problems. Great technology empowers people.
— Bjarni Kristján Leifsson

BREAK IT

Click anything. Watch what the rest of it does.

A production system, running. Every dot is a request being routed to a healthy instance — kill one and the survivors carry it, kill the last one and you find out what actually depended on it. This is the job, and it is more instructive than a bullet point saying “high availability”.

REQ/S
0
ERRORS
0%
p99
0ms
AVAILABILITY
100%
  1. No incidents. Click something to change that.

Research & experiments

The work behind the work.

Smaller pieces, but they explain where the systems thinking comes from.

  • MSc research

    Planning Support System for UX Evaluation

    Reykjavík University

    How structured UX-evaluation information can support planning and decision-making during software development. A framework and tool built around the factors that most affect evaluation planning, reducing the time and cost of an evaluation and helping manage product complexity.

    UX researchHCIRequirementsResearch methodsDecision support
  • Machine learning

    Predicting Working Memory from Cardiovascular Data

    Reykjavík University

    Machine-learning analysis of cardiovascular measurements from 98 participants performing increasingly stressful cognitive tests. A TreeBagger model reached 74% accuracy classifying which Stroop-test condition a participant was performing; the analysis also explored the relationship between time pressure, heart rate and blood pressure.

    Machine learningMATLABClassificationStatisticsData analysis
  • Data modeling

    M3 — Multidimensional Media Model

    Reykjavík University

    Applying multidimensional-analysis concepts to media data, including a prototype media server implementing the model for interactive exploration of multimedia collections. The M3 research line is still in use in Reykjavík University research today.

    Data modelingDatabasesOLAPMedia systemsResearch

Ideas & initiative

Work nobody asked for.

Concepts pursued outside any job or client, because the problem was worth solving.

  • Independent product proposal

    Apple Medical ID

    A product concept for Apple's Medical ID: getting blood type and emergency contact details off a locked phone and in front of a paramedic in seconds, for the case where the person it belongs to is unconscious or badly injured and cannot tell anyone anything.

    The problem is easy to state and it happens every day: someone is unconscious after an accident, the information that would help the people treating them — blood type, allergies, who to call — is on the phone in their pocket, and the phone is locked. Every second spent working around that is a second not spent on the patient. I researched who inside Apple owned that surface and wrote it up as a structured product document rather than a suggestion: the user story, the flow from lock screen to critical details, the interface and how it could be built. Then I sent it to the people who could act on it, and asked for nothing in return. Apple later shipped this functionality.

    This was my proposal to Apple. Whether it played any part in what they shipped, I have no idea.

    What it demonstrates

    Product thinkingSafety-critical UXUX designResearchProblem framingExecutive communicationInitiative