SDS project archive / 2025—26

WORK/

A working index of systems, prototypes, and research products. Each entry begins with the problem—not the technology.

5 active records
  1. SDS-01AML Risk AnalyticsFinancial complianceDelivered · 40% fewer false-positive alerts
  2. SDS-02CivilBotApplied AIIn development · Legal information, made legible
  3. SDS-03MediMeshHealthcare systemsIn development · FHIR-aligned by design
  4. SDS-04Telecom Churn AnalyticsDecision analyticsDelivered · From prediction to intervention
  5. SDS-05Sector Intelligence ProductsResearch & advisoryOngoing · Research structured for action
SDS-01Delivered

Financial compliance / Financial services

AML Risk Analytics

40% fewer false-positive alerts

A transaction-monitoring and anomaly-detection system built for auditability, faster review, and clearer regulatory reporting.

01 / Problem

A financial institution needed to detect suspicious transactions and meet regulatory reporting obligations. Manual review and basic rule-based alerts created noise while missing complex patterns.

02 / System response

SDS designed a risk-scoring system that combines statistical anomaly detection with configurable rules, reviewer workflows, and an auditable reporting layer.

Architecture / method

  • 01Real-time transaction ingestion
  • 02Statistical anomaly detection
  • 03Configurable risk-scoring engine
  • 04Compliance dashboard and audit trail

Value / direction

False-positive alerts fell by 40%, regulatory reporting moved from days to a repeatable workflow, and examiners gained a complete decision trail.

Record 01 / 05Back to index ↑
SDS-02In development

Applied AI / Legal / public interest

CivilBot

Legal information, made legible

A public-interest AI assistant that turns scattered civic information into clear, contextual guidance through natural language.

01 / Problem

Citizens struggle to navigate civic and legal processes because useful information is scattered, procedural, and written for specialists.

02 / System response

CivilBot pairs a curated legal knowledge base with retrieval-augmented generation, contextual conversation, and explicit feedback loops for safer iteration.

Architecture / method

  • 01Retrieval-augmented generation
  • 02Curated legal knowledge base
  • 03Context-aware conversation
  • 04Human feedback and evaluation loop

Value / direction

The system is designed to reduce information asymmetry by giving citizens a clearer starting point before professional legal support is required.

Record 02 / 05Back to index ↑
SDS-03In development

Healthcare systems / Distributed clinics

MediMesh

FHIR-aligned by design

A distributed health-data layer connecting clinics without flattening the governance, privacy, and operational realities between them.

01 / Problem

Clinics in distributed health systems operate in silos. Patient data stays locked at facility level, limiting care coordination, referrals, and system-wide reporting.

02 / System response

MediMesh creates facility-level nodes with a shared analytics layer, standardised data contracts, and role-based access that respects local control.

Architecture / method

  • 01Distributed facility-level nodes
  • 02HL7 FHIR-aligned data models
  • 03Cross-facility analytics layer
  • 04Role-based access and audit logging

Value / direction

The platform is being built to improve referral visibility, care coordination, and system-level decision support while preserving patient privacy.

Record 03 / 05Back to index ↑
SDS-04Delivered

Decision analytics / Telecommunications

Telecom Churn Analytics

From prediction to intervention

A churn intelligence system that connects subscriber behaviour, network quality, and retention action—not just another probability score.

01 / Problem

A telecom operator was losing subscribers without a reliable view of churn drivers. Retention spend was broad, reactive, and disconnected from service quality.

02 / System response

SDS combined usage, CRM, and network signals to identify at-risk segments, explain the drivers behind risk, and guide targeted retention decisions.

Architecture / method

  • 01CDR and CRM data integration
  • 02Churn prediction models
  • 03Customer segmentation analysis
  • 04Retention recommendation layer

Value / direction

The operator gained an actionable view of churn drivers and could intervene with the right customer groups before departure.

Record 04 / 05Back to index ↑
SDS-05Ongoing

Research & advisory / East African markets

Sector Intelligence Products

Research structured for action

Decision-oriented research products combining local datasets, primary inquiry, and market context for institutions operating in East Africa.

01 / Problem

Decision-makers often rely on sector reports that are outdated, generic, or detached from the operational realities of East African markets.

02 / System response

SDS combines quantitative analysis, expert interviews, and institutional context, then structures the output around the decision a reader needs to make.

Architecture / method

  • 01Primary and secondary research
  • 02Local quantitative datasets
  • 03Expert qualitative validation
  • 04Decision-oriented report frameworks

Value / direction

Investors, NGOs, and public institutions use the resulting intelligence to shape market entry, programme design, and policy decisions.

Record 05 / 05Back to index ↑

Your problem could be record 06

Give us the difficult version.

Start a conversation