ColdOps_re+start

Inspiration

Food and beverage factories in Malaysia lose tens of thousands of Ringgit every month because of hidden energy waste in their industrial cooling systems. While building software for clients like Marigold, we noticed that large cooling compressors kept running at full power even when production lines had stopped or storage rooms were completely empty. We joined the Taylor's University hackathon under the Smarter Resource Management track to solve this problem — turning invisible electricity waste into real cost savings for companies.

What It Does

ColdOps connects to existing Warehouse Management Systems (WMS) and uses that data to control how much electricity is sent to industrial coolers. It looks at the current time, how much stock is stored, and where items are placed across different cooling zones — then adjusts the cooling power accordingly. Before making any temperature changes, ColdOps always asks a human to review and approve the action first.

How We Built It

Frontend: Built with Next.js 14 (App Router) and TypeScript, featuring data visualizations and real-time savings counters powered by Socket.io. Backend & Database: Powered by Node.js with a custom rules engine, BullMQ and Redis for job queuing, PostgreSQL and TimescaleDB for time-series data storage, and Python for data processing.

Challenges We Ran Into

Without access to a real factory environment, we had no live data to test against. We had to manually simulate realistic warehouse condition, such as fluctuating stock levels, shift schedules, stock types, and temperature curves. Getting these numbers to feel believable enough to meaningfully stress-test our rules engine was harder than expected. Too clean and the system looked untested, while too random and the optimizations stopped making sense. We spent significant time researching real cold chain operation patterns just to build a simulation that reflected how an actual F&B facility behaves.

Accomplishments We're Proud Of

We successfully built an enterprise-ready system that requires zero upfront cost from the client, because it plugs directly into data streams the factory already has — no new hardware, no new infrastructure.

What We Learned

We learned that a purely technical optimization model fails in industrial settings unless it genuinely accounts for human workflows and strict safety regulations. Even when AI can handle a task confidently, keeping a human in the loop is not optional — it is a core requirement in real factory environments.

What's Next for ColdOps

Our immediate next step is deploying ColdOps as a fully tested, production-ready product across Double Dot's existing enterprise client base. We plan to launch our first live trial inside Marigold's processing facilities.

Built With

• javascript
• next
• postgresql
• python
• render
• timescale
• typescript