Asian Development Bank
National Food Security Information System
May 2023 to Dec 2023
Project category
Applied artificial intelligence and decision analytics
Acrux builds generative AI assistants, conversational applications, and analytics platforms that turn complex public sector and enterprise data into decisions people can act on.
Overview
From raw data to intelligent, explainable decision support
Our AI and analytics portfolio pairs large language models, natural language processing, and modern data engineering with strong domain understanding in trade, taxation, food security, and sustainability.
Delivered work includes the National Food Security Information System for the Asian Development Bank, a conversational application for electronic filing of income tax returns, the Trade Data Analytics System, and the NetZero Reality carbon accounting application.
Generative AILLMsConversational AIBusiness IntelligenceDecision Support
LLM models
AI assistants
Forecasting
Dashboards
Data pipelines
Insights
Capabilities
Where AI creates measurable value in our engagements
Conversational applications
Assistants that guide citizens and staff through complex filings and procedures in plain language.
Large language model integration
Retrieval grounded generation over official documents, rules, and organizational knowledge.
Document intelligence
Extraction and classification of structured facts from unstructured reports and forms.
Analytics and forecasting
Trend, seasonality, and scenario analysis over trade, price, production, and consumption data.
Interactive dashboards
Superset and Highcharts based dashboards for executives, analysts, and field teams.
Data engineering
Ingestion, cleansing, and harmonization pipelines across fragmented public data sources.
Risk analytics
Scoring and profiling models that surface anomalies and compliance risk early.
Monitoring platforms
Indicator frameworks with alerts for food security, trade performance, and emissions.
Responsible AI practices
Human review, traceable sources, and access controls built into every AI workflow.
Selected engagements
Delivered for governments, development partners, and enterprises
Federal Board of Revenue
Conversational App for E filing of Income Tax Returns
2022
Public sector
Trade Data Analytics System
2022
Sustainability partner
NetZero Reality App for Carbon Accounting
2025
Product
PostJin, AI generated social media content management platform
2024
Delivery approach
How we take an AI initiative from idea to production
- Step 1
Use case discovery and value assessment with domain owners.
- Step 2
Data inventory, quality review, and access design.
- Step 3
Prototype with real data and measurable success criteria.
- Step 4
Model selection, prompt design, and retrieval grounding.
- Step 5
Human in the loop review and validation workflows.
- Step 6
Integration with existing systems and dashboards.
- Step 7
Deployment with monitoring, logging, and cost controls.
- Step 8
Continuous evaluation and iteration on live feedback.
Technology highlights
Engineering foundations we rely on
- Retrieval augmented generation over trusted sources
- Natural language interfaces for complex procedures
- Automated indicator and KPI computation
- Interactive visualization at national scale
- Traceable outputs with source references
- Cloud native, cost aware AI deployment
Business benefits
Outcomes our clients measure
- Faster access to insight without specialist analysts.
- Reduced effort in preparing routine filings and reports.
- Earlier detection of anomalies and emerging risks.
- Consistent, explainable answers grounded in official data.
- Better coordinated decisions across agencies and teams.
- Higher analyst productivity through automation.
Technologies used
A production ready stack from frontend to infrastructure
AI
Large Language ModelsHugging FaceNatural Language ProcessingSpeech to Text
Data and Analytics
Apache SupersetHighchartsPythonSQLMongoDB
Applications
React.jsNode.jsFastAPIREST APIs
Infrastructure
AWSDockerMonitoringRole Based Auth
Have an AI or analytics use case in mind
Share your data and decision challenge and we will map a practical path from prototype to production.
Talk to our team