Data platforms and AI on AWS and Google Cloud
What we build
| Layer | What we build | What you decide |
|---|---|---|
| Ingestion | Pipelines from operational systems and streams, with data-quality checks | Which sources are in scope, and who owns each one |
| Warehouse | A governed cloud warehouse on Google BigQuery or an AWS-based warehouse | Retention, data location and access rules |
| Dashboards | Self-serve BI on Power BI or Looker, or on BizInsight | Metric definitions |
| Machine learning | Training, serving and MLOps on Vertex AI or Amazon SageMaker | Which use cases justify a model, and the acceptance criteria |
| Operations | Monitoring, access control, cost guardrails | Who approves changes |
• BizInsight. A TiCON business-intelligence platform for dashboards, custom reporting and data integration from sources such as CRM and ERP systems. It includes access controls, encryption and audit-ready tracking. Cloud work runs on AWS where the customer's stack is AWS.
• LixA. A TiCON AI and data-science platform for predictive analytics, decision automation, natural language processing, computer vision and data pipelines. We do not publish accuracy or ROI figures for LixA; any figure for your data should come from your own evaluation.
A telecom example, and Korea-market work
Banglalink, a telecom operator, had data spread across systems and reporting that took days. TiCON built a cloud warehouse, streaming pipelines and self-serve dashboards, with access controls and cost guardrails. Turnaround moved from days to minutes, about ten times faster. The case reports a petabyte-scale warehouse, and its figures are representative of TiCON engagements and may be anonymized. Read the case study
Our Korea-market work includes a smart-data project for Busan Metro, a metro operator with 1.5M passengers a day, and sensor and alarm work for GSpider, a Korean Army program. Korea-market references are listed on the about page.
Where the data sits, and how we choose the cloud
Bangladesh has no hyperscaler region, so a data platform built here runs in another country. Our working regions are AWS Mumbai (ap-south-1) and Singapore (ap-southeast-1), and Google Cloud asia-south1 (Mumbai), asia-south2 (Delhi) and asia-southeast1 (Singapore). Put the region decision into the design before the build starts. If a category of data must stay in one country, we design a hybrid or local path and write it down. The data residency page explains how we approach that decision.
Start from the product you will run. Google publishes BigQuery and Vertex AI as strengths of its platform, and AWS offers a broad service catalogue. Keep each application where it runs best, and move only the data or ML workload when that is the better fit. Two identity systems and two bills with no clear reason become expensive quickly. The cloud choice guide sets out the criteria.
How an engagement runs
- Step 1:
Define the goal and the sources.
Agree the business questions, the data sources and who owns each source.
- Step 2:
Design and build.
Pipelines, models and dashboards, built against the agreed questions.
- Step 3:
Test and deploy.
Data-quality and accuracy checks run before users see a number.
- Step 4:
Monitor and tune.
Retrain models, tune pipelines and keep costs under watch.
What you still own: the business definition of every metric, data ownership, retention and location decisions, model acceptance criteria and who signs them off, and your identity provider and access approvals.
After go-live
A data platform runs like any production system. Monitoring and on-call for in-scope infrastructure run 24x7, and managed engagements include a named Technical Account Manager. A monthly cost review covers warehouse and compute spend. Retraining and pipeline changes need approval from the owner you named at the start.
Cost
Data platform work is quoted per scope, in USD. Compute and storage used by the platform are billed under your cloud agreement, or under partner billing where a contract includes it. Where partner billing applies, compute and storage pricing has typically come in 15 to 35 percent below list. That range is an estimate, and actual results depend on your workload. Other currencies are available only by contract. Taxes apply as applicable in your jurisdiction and the contract.
This page does not claim:
Accuracy, ROI or productivity figures for LixA
SLA minutes, which sit in the contract
Named client projects for our Vertex AI and Amazon SageMaker work. We describe that work as capability.
This page is not legal advice. Confirm applicable requirements with your counsel and compliance team.