Projects spanning startup delivery, cloud architecture, cybersecurity infrastructure, research workflows, aviation recovery field operations, Arctic expedition planning, and audio post-production.
Gale Force North works where technical systems meet operational reality: production environments, research workflows, field constraints, and teams that need dependable execution. These selected projects show the type of judgment, structure, and ownership brought to complex work.
The Founder page gives the background. The examples below show the same operating discipline in specific project situations.
Projects where structure, ownership, and execution mattered.
The examples below are selected work where unclear conditions had to become an operating path: product architecture, infrastructure, workflow, field logistics, research systems, audio post-production, or delivery recovery.
Selected Gale Force North client engagement with a Rannís-funded Icelandic startup, bringing a service-marketplace concept into a structured product, architecture, cloud platform, design handoff, and delivery path.
The project is a two-sided marketplace for Icelandic licensed trade work, where product design, trust, scope definition, offers, agreements, documents, payments, and property maintenance records all have to become one coherent software system. The risk is not only technical. Without clear architecture and delivery structure, the product can drift into scattered screens, informal decisions, weak trust signals, unclear payment handling, and infrastructure that is hard for future contributors to understand.
Role
Through Gale Force North, Arnar Ingi provides technical leadership across product architecture, implementation direction, cloud infrastructure, delivery planning, UI/UX hiring, design handoff, and coordination with external design work for this client engagement. He has worked with the project over an extended period and has carried the technical foundation into structured delivery artifacts and cloud-platform implementation.
Approach
The work connects product intent to implementation. A system architecture and infrastructure plan defines the frontend, backend API, database, identity, environments, deployment, monitoring, backup, and security model. A marketplace design brief translates the product into roles, journeys, requirements, trust signals, payment states, evidence records, and development-ready design-system expectations. The cloud work has established a governed AWS foundation with development, staging, production, and shared networking accounts, Terraform/Terragrunt workflows, DNS and certificate controls, operator-access boundaries, and repeatable delivery practices. Design coordination includes hiring for UI/UX capability and preparing external design work so it can be reviewed and built against explicit requirements rather than guesswork.
Outcome
The project has gained a much clearer operating basis: documented product scope, architecture direction, implementation-ready UI flows, design-system expectations, a cloud infrastructure plan, a multi-environment AWS footing, DNS and certificate control model, and a delivery path that future contributors can inspect. The work has not merely added isolated code or design input; it has turned a broad marketplace idea into a structured product and platform program.
Selected Work 02
Frostbyte / NCC-IS Lab Buildout
Year-long work driving the Frostbyte / NCC-IS lab buildout from concept and limited infrastructure into a working technical environment with servers, networking, virtualization, storage, student teams, and operational structure.
Launching Reykjavík University's part of the Frostbyte / NCC-IS lab required more than a nominal lab concept. The project needed physical infrastructure, server-room installation, virtualization, networking, Linux environments, internal services, backup strategy, student coordination, web presence, and a working space where technical development could happen.
Role
Drove the project forward across technical leadership, procurement, physical setup, student hiring, student supervision, infrastructure planning, and operational execution through a year-long launch and early-operations period.
Approach
The work followed a startup-style execution model: secure and equip a working space, assemble a motivated student team, define the technical direction, and build usable infrastructure quickly. Physical servers were procured and installed, networking was planned around MikroTik equipment, Proxmox provided the virtualization layer, Linux virtual machines supported lab services and exercises, and internal tooling included GitLab plus a MinIO-based backup strategy.
Outcome
The project moved from limited initial footing to a functioning Frostbyte / NCC-IS lab environment with physical servers, virtualization, networking, Linux systems, internal development tooling, backup strategy, web presence, student participation, and operational direction. It helped create the foundation for a lab that later gained dedicated workspace and long-term research and education value.
Selected Work 03
University Programming Teaching
University teaching work at Reykjavík University across first-year programming, data structures, practical Python project architecture, web programming, grading, and student support.
University teachingPythonData structuresGitHubProject architectureReactAPIsGrading
Challenge
Early programming students need to move from isolated exercises into structured software work without losing the fundamentals. The teaching model at Reykjavík University put professor-owned courses into smaller semester groups, where each assigned teacher was responsible for guiding a cohort through regular sessions, exercises, grading, and student support.
Role
Worked as a cohort-responsible university teacher within professor-owned first-year programming and data-structures courses, leading assigned groups of roughly 50 students through twice-weekly four-hour sessions across the semester. Also supported related practical and web-programming courses, including work as the sole main teaching assistant for Vefforritun 2 over several semesters, grading roughly 120 students weekly.
Approach
The work combined direct instruction, practical exercises, grading, student support, and local teaching decisions inside the course structure. A major part of the programming pathway was the intensive three-week practical Python project, where student teams learned to build larger applications with GitHub, separation of concerns, classes, a text UI, a logic layer, and CSV/JSON-backed data storage. Later web-programming support covered React, vanilla JavaScript, more advanced HTML/CSS, backend fundamentals, and APIs.
Outcome
The teaching work helped students cross the gap from syntax and small assignments into team-based project structure, layered application design, version-control practice, and web-application fundamentals. For Gale Force North, it is evidence of technical leadership through explanation, assessment, architecture framing, and sustained responsibility for helping groups work inside complex material.
Selected Work 04
Memory Forensics And OS Identification
Applied MSc cybersecurity research using deep learning to identify operating systems from memory dumps, with kernel extraction, binary disassembly, Slurm-based HPC processing, and GPU-accelerated training.
Operating-system identification from memory snapshots is useful for VM management, security monitoring, and forensic analysis, but raw memory dumps are large, noisy, and difficult to compare directly. The project needed a scalable pipeline that could turn kernel-level binary material into structured features suitable for machine-learning classification.
Role
Designed and implemented the research pipeline for an MSc cybersecurity project, covering data preparation, kernel extraction, binary disassembly, feature-vector generation, HPC processing, and GPU-accelerated model training.
Approach
The workflow started from a 1.5 TB memory-dump dataset and reduced it to a 39 GB kernel dataset before transfer to the HPC platform. Feature extraction used the iced_x86 library for kernel disassembly, histogram-based function-size features, and Slurm array processing across 100 CPU tasks, reducing a roughly eight-hour serial job to about twenty minutes. A GPU-trained feedforward neural network then classified the retained operating-system classes from the processed vectors.
Outcome
The project produced a working applied-research pipeline across cybersecurity, binary analysis, data processing, HPC orchestration, and deep-learning model training, reaching 99.42% accuracy across the retained classes while still showing limits around rare classes. For Gale Force North, it is evidence of practical work at the boundary between software, infrastructure, research data, and security analysis.
Selected Work 05
Greenland Ice Sheet Aviation Recovery & Under Ice
Field participation in the Air France Flight 66 fan-hub recovery on the Greenland Ice Sheet, followed by the four-month creation of Under Ice, an IMDb-listed documentary artifact about the mission.
Aviation recoveryGreenland Ice SheetSARArctic field operationsTechnical extractionDocumentary production
Challenge
Recovering an aircraft engine component from the Greenland Ice Sheet is not a normal field task. The missing fan hub fragment had separated from an Air France A380 during flight and was needed for further aviation safety investigation. After nearly two years of scientific search and localization work by specialist teams, the recovery party was flown to the confirmed site in a remote crevasse-field environment where the component was buried under snow and ice.
Role
Contributed to the Icelandic SAR field effort supporting the recovery operation, bringing technical rescue experience, glacial field judgment, practical extraction capability, and operational discipline to work where safety, precision, and execution mattered. After the mission, Arnar Ingi created Under Ice, a 30-minute documentary about the final recovery expedition, over a four-month production period. The film is listed on IMDb and serves as a public documentary artifact for the project.
Approach
The recovery was treated as a controlled field operation. Crevasses, snow bridges, weather, polar-bear risk, camp safety, heavy-object handling, and helicopter-supported logistics all had to be handled as operational constraints. The work involved crevasse-area safety, fall-arrest systems, digging and melting through snow and ice, multi-directional rope hauling, sled movement, winch support, and preparing the more than 200 kg fragment for helicopter sling transport. The documentary work then turned the field material into a coherent account of the mission context, camp work, extraction process, crevasse risk, rope systems, and helicopter sling preparation.
Outcome
The fan hub fragment was recovered from the Greenland Ice Sheet and delivered for further inspection as part of the aviation safety investigation. Under Ice preserved the final recovery work as a public documentary record, extending the value of the field operation beyond the extraction itself.
Under Ice Trailer
One-minute trailer for Under Ice, Arnar Ingi's documentary about the Greenland ice-sheet recovery of the Air France Flight 66 fan-hub fragment.
Loading video...
Greenland Ice-Sheet Recovery Site
Field camp and marked work area on the Greenland ice sheet during the fan-hub recovery mission.
Loading video...
Under Ice Documentary
The full 30-minute documentary created over a four-month production period, following the final recovery operation on the Greenland ice sheet.
Selected Work 06
Hálfur Álfur: Audio Reconstruction And Post Production For Cinema Delivery
Audio post-production, restoration, mixing, and mastering for a documentary film, turning a difficult handoff into a structured, cinema-ready Pro Tools session.
The production arrived after a difficult handoff with audio material that was roughly chronological, but structurally unsuitable for professional post-production. The film needed a coherent audio workflow: organized source material, usable dialogue tracks, controlled restoration, foley where needed, and a final mix and master that could support cinema delivery.
Role
Led the audio post-production, drawing on years of recording, mixing, mastering, and music-production experience. The work covered session reconstruction, dialogue editing, restoration, mix preparation, final mastering, and a custom Python workflow to rename and reorganize source audio for import into Pro Tools.
Approach
The first step was to impose order on the source material. Python automation helped rename and arrange files for controlled transfer into Pro Tools. The session was then reconstructed manually and editorially: dialogue was grouped, ambience separated, music placed into appropriate tracks, and supporting material organized into a usable mix architecture. Restoration work used iZotope RX, tonal shaping used tools such as FabFilter where appropriate, and foley was added where needed for continuity and scene clarity.
Outcome
The project moved from a difficult-to-use audio handoff to a well-structured, restored, mixed, and mastered cinema-ready delivery. The finished film supported a public release path across festivals, Icelandic broadcast, and later streaming distribution. Hálfur Álfur won the Skjaldborg Grand Jury Prize, received an Edda Award nomination, screened internationally, aired on RÚV, and has appeared on Netflix.
Selected Work 07
Subsea Fibre Seismic Research
Ongoing MSc thesis research using high-performance computing to compare earthquake simulations with observed signals on the Iceland-Ireland subsea cable, presented as a research poster at EGU 2025.
EGU 2025 posterNokia Bell Labs collaborationResearch infrastructureHPCEarthquake simulationsSubsea cable dataPython
Challenge
The research sits between geophysics, signal processing, scientific computing, and infrastructure. It uses data from the Iceland-Ireland subsea cable and compares observed cable signals with synthetic earthquake waveforms, which requires careful data handling, simulation workflow design, high-performance computing, and enough structure for research collaborators to inspect and extend the work.
Role
Carrying out ongoing MSc thesis research after completing the coursework component of the degree, with collaboration involving Nokia Bell Labs and academic partners. Presented the work as a research poster at EGU 2025, the European Geosciences Union General Assembly, Europe's largest and most prominent geosciences event.
Approach
The work uses high-performance computing platforms to run earthquake simulations and compare synthetic waveforms with observed subsea-cable signals. The implementation focus is reproducible research workflow: organized data processing, repeatable simulation runs, structured comparison of observed and synthetic signals, and code that remains understandable as the research question evolves.
Outcome
The project has produced a visible research artifact through the EGU 2025 poster and continues as ongoing thesis work. For Gale Force North, it is evidence of applied technical work across scientific data, HPC, signal processing, research infrastructure, and external collaboration.
EGU 2025 Research Poster
Research poster presented at EGU 2025, the European Geosciences Union General Assembly, for ongoing MSc thesis work on seismic waveforms observed on the Iceland-Ireland subsea cable.
Selected Work 08
Svalbard SAR Expedition Planning & Execution
Planning and executing a 60 km Arctic expedition for 18 members of an Icelandic search-and-rescue team, coordinating route design, polar-bear risk, weather, people, equipment, authorities, accommodation, transport, and marine extraction.
Moving 18 people safely through 60 km of Svalbard terrain required more than choosing a route. The expedition involved Arctic weather, rough terrain, polar-bear risk, communications constraints, group fatigue, accommodation, transport, equipment, authority coordination, and the practical reality of keeping a large volunteer team aligned over months of preparation.
Role
Planned and executed the expedition for members of an Icelandic search-and-rescue team. Led route planning, participant coordination, deadline management, travel logistics, accommodation coordination, safety-equipment planning, authority coordination, and extraction planning.
Approach
The expedition was planned as a controlled field operation. The route was designed around distance, terrain, weather exposure, group capability, camp logistics, and extraction feasibility. Safety planning included tent-group organization, night-watch structure, trained rifle capability in each tent group, flare guns, satellite phones, and coordination with Svalbard authorities and local operators. Marine extraction from Barentsburg was arranged with RIB boats and support boat capacity from Longyearbyen.
Outcome
The expedition was planned, coordinated, and executed as a successful 60 km Arctic field operation for 18 participants, with a defined route, safety structure, communications plan, polar-bear risk controls, camp organization, and planned marine extraction.
Svalbard Expedition Video
Field video from the three-day, 60 km search-and-rescue training expedition.
Loading video...
Selected Work 09
Booking Platform Product R&D
Earlier two-year product-platform work on a tour and booking system, covering Next.js application development, Redux cart state, checkout flow, Prisma/PostgreSQL models, API routes, Prismic content rendering, authentication, and payment-integration experiments.
A serious booking platform pulls many concerns into one system: product pages, schedules, participant counts, carts, contact details, checkout stages, payment redirection, content management, authentication, admin tools, and data persistence. The work showed how quickly frontend state, content, database shape, APIs, payments, auth, and infrastructure become hard to reason about when they are not owned as one operating path.
Role
Designed and implemented substantial parts of an independent booking-platform effort over roughly two years. The product was not completed into a finished commercial platform, but the work became practical product-platform R&D that shaped how Arnar Ingi thinks about architecture, delivery boundaries, state, content, checkout, and infrastructure ownership.
Approach
The implementation used a Next.js and React application with Chakra UI, Redux Toolkit, Next API routes, Prisma over PostgreSQL, Prismic-driven page rendering, JWT/Cognito-style authentication utilities, cart and registration APIs, checkout-stage pages, and Rapyd payment-integration experiments. The platform included booking-panel UI, date and participant selection, cart state, registrant/contact data, schedule and engagement models, management pages, publishing tools, user-management surfaces, and a custom article-builder direction.
Outcome
The work did not become a finished booking product, but it produced a large amount of concrete implementation experience across product architecture, frontend state, CMS content, data modeling, server-side API boundaries, checkout, payments, authentication, and maintainability. It directly informed the Whirlwind emphasis on owning frontend, backend, content, auth, deployment, and infrastructure decisions together before a product grows around improvised choices.
Next Step
Have a project that needs structure, ownership, and technical direction?
Gale Force North helps teams turn uncertain technical work into controlled delivery: architecture, infrastructure, deployment, environments, access, security, workflow, logistics, and operational process.
