Smart HVAC Optimisation – Scaling a cloud-based IoT platform
A Norwegian provider of cloud-based HVAC optimisation solutions partnered with DO OK to improve the scalability, performance and operational efficiency of their IoT platform. Over the course of the engagement we eliminated major technical bottlenecks, reduced infrastructure costs, improved data governance and introduced customer-facing features that made the platform more valuable for building managers. The result was a more reliable, cost-effective system that supported the client’s growth from startup to scale-up.
The Client
Our client is a Norwegian technology company specialising in automated and optimised control of ventilation systems for commercial buildings. Their mission is to help customers maintain a healthy indoor climate while reducing energy consumption and CO₂ emissions.
Their cloud-based platform optimises HVAC systems in real time, drawing on occupancy sensors, indoor air quality measurements and weather forecasts. By adjusting ventilation rates dynamically, the system can reduce energy use and emissions by 15–25% while keeping CO₂ levels and temperature stable.
The solution integrates with or without Building Management Systems (BMS) through a proprietary IoT gateway that supports multiple communication protocols. Users, typically building managers for large-format properties such as warehouses, factories and retail spaces, access a dashboard to monitor energy savings, indoor conditions and portfolio-wide performance metrics.
Languages
TypeScript
HTML
JavaScript
CSS
Frameworks
NestJS
Node.js
TypeOrm
Dynamoose
Serverless Framework
Angular
Other Tools
AWS
PostgreSQL
DynamoDB
Cloudformation
Elastic Container Service
Balena Cloud,
Docker
IoT Hub
REST
Project Size
3 x Fullstack Developers (incl. DevOps)
+5500 hours
Client Needs
The client approached DO OK seeking a trusted, technically strong partner to stabilise and evolve their cloud-based SaaS IoT platform, a system critical to their mission of helping building managers cut energy use and emissions. Their in-house development team had dispersed, creating both operational and strategic risk. In the past different areas of the platform, from integrations to infrastructure, had been handled by multiple specialists. Now they needed a single accountable partner who addressed immediate technical challenges and provided long-term advisory support.
The platform faced growing pains that directly affected the business. High cloud infrastructure costs were eroding margins, onboarding new buildings took up to three months and fragile integrations with external data providers risked service interruptions. Data inconsistencies and delays in detecting faults reduced user confidence, while the lack of a clear process for releases sometimes caused downtime.
A key example was excessive microservice chattiness, where services generated a large volume of internal calls. This not only inflated operational costs but slowed the flow of telemetry data, meaning customers saw outdated information in their dashboards. Combined with loosely typed implementations and an underdeveloped API, these issues impacted both the user experience and the client’s ability to scale.
They were looking for a partner who acted independently, resolved long-standing inefficiencies, advised on product direction and delivered changes quickly without heavy oversight, ultimately enabling the platform to support more customers, more buildings and more integrations reliably.
Why DO OK?
The client initially hesitated to work with a software house, having only partnered with in-house development teams before. They feared an external provider might lack the agility, business understanding and hands-on ownership they were used to. What shifted their perspective was DO OK’s proven track record in complex cloud systems and our ability to advise on both technical and product-level decisions.
Before starting delivery our team spent two weeks on site in Norway, working closely with the client’s CTO, product owner, CEO and other key stakeholders. This period combined technical discovery with a deep dive into their business context, ensuring we understood not only the platform’s architecture but also how it was used in real-world operations. We met with representatives from multiple departments, observed workflows and identified both quick wins and strategic priorities.
By the end of onboarding we had already completed work the client expected to take months, an early sign of our efficiency and ability to work independently. This built trust quickly and reassured them that DO OK maintained continuity while pushing for improvement.
Our combination of strong technical capability, proactive problem solving and business-focused advice ultimately reinforced their decision. This trust was further validated when, based on our performance, the client recommended DO OK to their parent company, with whom we went on to deliver another project.
Project Description
The engagement began with a full audit of the IoT platform, assessing architecture, integrations and infrastructure costs. This included reviewing how the system performed under real-world conditions and gathering input from operational teams to identify recurring issues and bottlenecks. The audit findings were presented to the client’s management team, outlining a clear roadmap for stabilisation and cost optimisation.
One of the earliest and most impactful actions was addressing microservice chattiness. The system’s 12 loosely coupled services generated excessive internal network calls, inflating operational costs and slowing the processing of telemetry data from buildings. Within less than two weeks we consolidated related services into larger domain-consistent units, reduced internal calls by around 70% and halved the number of services to 6. This immediately improved platform responsiveness, cut monthly cloud costs and simplified maintenance.
In parallel, we worked with the product owner to resolve domain-specific challenges. For example, historical data from certain buildings lacked standardised formatting, which distorted the analytics presented in the dashboard. We improved data governance by introducing stricter validation rules and consistent handling throughout the pipeline. This reduced the need for manual checks and improved trust in the platform’s output.
We also expanded the platform’s compatibility through new integrations. Weather service data was integrated to improve predictive ventilation adjustments, building management systems were connected to enable direct control, and new energy metering provider links were developed to support more diverse customer profiles. One customer-facing feature, the Info Screen, was designed for retail stores to display real-time indoor air quality and energy savings to staff and customers.
Technical debt was reduced steadily throughout the engagement. We replaced unstructured JavaScript with strictly typed TypeScript, aligned code with clean architecture principles and redesigned the adapter architecture so that all building controllers communicated through a unified entry point. This eliminated previous cases where faulty integrations either failed silently or produced inconsistent results, which had been a source of customer complaints.
The project was delivered in an agile, long-term support model. We alternated between infrastructure optimisation, feature delivery and process improvements, ensuring that immediate operational needs were met without losing sight of the platform’s long-term scalability. The ability to move independently between fixing infrastructure, implementing new integrations and advising on product direction was key to sustaining progress and creating a system capable of supporting the client’s growth.
Key Challenges We Solved
During the engagement we addressed several issues that were holding the platform back. In larger buildings with high sensor counts, the cloud database was reaching storage limits, risking data loss and requiring manual oversight. We optimised storage management to handle larger volumes efficiently, eliminating the need for constant monitoring.
The process for ensuring data quality was also highly manual, with technical staff running hundreds of SQL queries daily to spot missing or inconsistent readings. We automated these checks and introduced real-time alerts, reducing operational overhead and allowing faster response to issues.
The release process posed another risk, as changes were tested directly in production, occasionally causing outages. We introduced a safer delivery workflow with staging and history management, ensuring stability before deployment.
Finally, integrations with building controllers were prone to failure due to inconsistencies in the adapter architecture. We redesigned it to use a single, unified entry point, reducing maintenance needs and preventing silent failures.
These improvements not only solved immediate operational problems but also removed long-standing technical constraints, giving the platform a stronger, more scalable foundation for future development.
Impact & Outcomes
During our collaboration the platform became significantly more stable, scalable and cost-efficient. Infrastructure optimisation reduced monthly operating costs, while consolidation of services improved response times and system uptime. Automated monitoring replaced manual daily checks, greatly improving data accuracy and enabling issues to be detected automatically before they were reported by customers. This allowed problems to be resolved more quickly and reduced the operational impact of data quality issues.
Operational efficiency improved markedly. Even though the client’s internal development team was smaller than before, the quality and speed of delivery did not drop. The platform became easier to maintain, freeing in-house resources to focus on strategic initiatives instead of firefighting operational problems. The simplified architecture also meant that new features could be delivered faster, with less risk of introducing regressions.
The redesigned adapter architecture strengthened integration resilience, significantly reducing downtime caused by external data provider changes. This in turn allowed the client to onboard larger and more complex buildings without fear of stability issues, expanding their addressable market.
Other measurable gains included a reduction of internal network calls by around 70%, halving the number of microservices, faster dashboard updates, improved scalability for future growth and shorter onboarding times for new buildings. Customers benefited from faster, more accurate data in the dashboard, while the business gained a stronger technical foundation for ongoing development.
