We design and build resilient computer systems rooted in sustainable architecture. From cloud-native infrastructure to edge computing networks, every solution we craft balances performance with environmental responsibility.
We engineer the invisible infrastructure that powers modern enterprise — from distributed cloud platforms to intelligent edge networks. Each engagement begins with your business cadence and scales outward from there.
Cloud Infrastructure Architecture
We design multi-region, auto-scaling cloud platforms on AWS, Azure, and GCP. Our reference architectures emphasize cost-aware resource scheduling, zero-trust networking, and infrastructure-as-code reproducibility. Every deployment includes chaos-engineering playbooks and mean-time-to-recovery targets below four minutes.
Kubernetes cluster design and fleet management
Serverless orchestration with step-function workflows
Multi-cloud disaster recovery with active-active failover
FinOps governance and reserved-instance optimization
Edge Computing & IoT Networks
From factory-floor sensor meshes to distributed retail analytics, we build edge networks that preprocess data at source and reduce cloud egress by up to 80%. Our reference stacks span MQTT brokers, Rust-based stream processors, and lightweight container runtimes on ARM edge nodes.
MQTT / OPC-UA protocol bridging and normalization
On-device inference with TensorFlow Lite and ONNX
Secure device provisioning at scale via PKI
Digital-twin synchronization with sub-second latency
Systems Integration & API Design
We connect legacy monoliths, SaaS platforms, and bespoke microservices into coherent data flows. Our integration practice covers gRPC service meshes, GraphQL federation layers, event-sourcing pipelines on Kafka, and ETL orchestration with dbt and Airflow. Every contract ships with an OpenAPI spec, a Postman collection, and a chaos-tested SLA.
Event-driven architectures with CQRS and event sourcing
GraphQL supergraph composition and schema governance
Legacy SOAP-to-REST migration with zero-downtime cutover
Real-time CDC pipelines from OLTP to OLAP stores
Performance Engineering
We profile, benchmark, and harden systems under production load. Our methodology combines flame-graph analysis, kernel-level tracing with eBPF, and statistical modeling of tail latencies. Typical engagements yield a 40–70% reduction in p99 response times while lowering compute spend through rightsizing and workload colocation.
JVM / .NET / Node.js runtime profiling and GC tuning
Database query-plan analysis and index optimization
Cache topology design with Redis, Memcached, and CDN layers
Load-test automation integrated into CI/CD pipelines
Security & Compliance Architecture
We embed security into the system design phase, not bolt it on afterward. Our architects model threat surfaces using STRIDE and attack trees, then instrument defenses from the kernel up: secure enclaves, mTLS service meshes, WAF rulesets, and runtime anomaly detection. We guide organizations through SOC 2 Type II, ISO 27001, and HIPAA readiness.
Zero-trust network architecture and microsegmentation
Secrets management with HashiCorp Vault and KMS hierarchies
SBOM generation and vulnerability lifecycle management
Incident-response runbook development and tabletop exercises
Technical Strategy & Advisory
For organizations navigating platform modernization, build-vs-buy decisions, or engineering-org scaling, we provide vendor-neutral technical due diligence. Deliverables include architecture decision records, capacity models, total-cost-of-ownership projections, and phased migration roadmaps with rollback checkpoints at every stage.
Technology radar and stack selection frameworks
Build-vs-buy analysis with quantitative scoring models
Engineering-team topology and Conway's Law assessment
RFP authoring and vendor evaluation scorecards
About the Studio
Where deep engineering meets ecological thinking
Green Horizon Studio was founded on a straightforward observation: the systems that run our world consume enormous amounts of energy, and most of that energy is wasted on inefficiency. We set out to prove that performance and sustainability are not competing priorities — they are the same priority, seen through different lenses.
We are a compact, senior-heavy team based in Hong Kong. Every architect on staff has at least twelve years of production systems experience across finance, logistics, health-tech, and telecommunications. We carry no junior bench, no account managers, no filler. When you engage us, the people in the room are the people who design and deliver the work.
Our practice is deliberately technology-agnostic. We select tools based on the problem, not the partnership. Over the years we have built deep capabilities across the AWS, Azure, and GCP ecosystems, the CNCF landscape, and a range of data platforms — but our real expertise lies in composing these pieces into systems that outlast the roadmap cycle.
Depth over breadth
We take on fewer engagements and go deeper on each one.
Long-term thinking
We design for the decade, not the quarter.
Radical candor
We tell clients what they need to hear, not what they want to hear.
12+years average seniority
8senior architects
1office, zero silos
Green Horizon Studio is a trade name of
ONN GLOBAL LIMITED
1/F 11E CHE HA VILLAGE
SAI SHA ROAD
Sai Sha, Hong Kong
Selected Work
Systems we have shipped
Representative engagements across industries. Names redacted where confidentiality agreements apply; architectural details shared with permission.
Logistics
Real-Time Fleet Orchestration Platform
Designed and delivered a streaming-data pipeline ingesting telemetry from 40,000 delivery vehicles across Southeast Asia. The system correlates GPS, engine-diagnostics, and route-plan data through a Kafka–Flink stream processor, feeding a digital-twin dashboard that dispatchers use to reroute vehicles in response to traffic anomalies. p99 processing latency dropped from 22 seconds to 340 milliseconds after our rearchitecture.
Architected a multi-petabyte data platform for a genomics research consortium spanning twelve hospitals. The solution layers AWS Lake Formation over S3 with Parquet-based storage, Apache Iceberg for time-travel queries, and a purpose-built de-identification pipeline that strips PHI before data enters the analytical zone. Query performance improved 8x over the legacy HDFS cluster while reducing storage costs by 62%.
AWS Lake FormationApache IcebergEMRAthenadbt
FinTech
Low-Latency Payment Switch Migration
Led the zero-downtime migration of a payment switch processing 1,800 transactions per second from an on-premises mainframe to a cloud-native architecture. The new system uses a Rust-based transaction core with an actor-model concurrency pattern, deployed across three availability zones with Raft-based consensus for ledger consistency. End-to-end authorization latency fell from 420ms to 18ms at the 99th percentile.
RustgRPCPostgreSQLConsulEnvoy Proxy
How We Work
A process shaped by a thousand projects
Our engagement model is designed to de-risk complex systems work. Every phase produces a concrete, reviewable artifact — no hand-waving, no black boxes.
01
Discovery & Context-Gathering
We embed with your engineering and product teams for one to two weeks. The output is a Systems Context Document: current-state architecture diagrams, a capability-gap analysis, a ranked risk register, and a prioritized backlog of architectural decisions that need to be made. We interview stakeholders across engineering, security, compliance, and finance to ensure no blind spots.
Deliverable: Systems Context Document, Risk Register, ADR Backlog
02
Architecture Design & Validation
We produce a Target Architecture Document covering the logical, process, and deployment views of the system (aligned with the 4+1 architectural view model). Each architectural decision is captured as an Architecture Decision Record with the context, options considered, trade-off analysis, and the rationale for the chosen approach. We validate critical paths with spike implementations and load-profile models before committing to the full build.
Deliverable: Target Architecture Document, ADRs, Spike Results, Cost Model
03
Incremental Build & Integration
We ship in thin vertical slices, each one a deployable, tested increment of the system. Our teams work inside your repositories, your CI/CD pipelines, and your code-review processes — we are not a black box that throws code over a wall. Every pull request includes design notes, test coverage reports, and an operations runbook entry. We pair with your engineers throughout so that institutional knowledge transfers in both directions.
We do not vanish after go-live. Our final phase runs for four to eight weeks post-deployment: we monitor system telemetry alongside your on-call team, tune alert thresholds to reduce noise, document incident-response procedures, and train your engineers on the operational surface of the system. The engagement closes with a retrospective and a prioritized backlog of suggested improvements — no ongoing dependency on us required.
Deliverable: Operations Runbook, Training Sessions, Improvement Backlog, Retrospective
Start the Conversation
Let us build something enduring
We take on a small number of engagements each year. If your project involves complex distributed systems, real-time data processing, or infrastructure at scale, we should talk. Reach out with a brief description of your challenge and we will respond within one business day.