Open source · inside the bank

Open-source infrastructure
and AI, deployed
inside the bank.

We run the open-source stack — databases, middleware, infrastructure — for 26 institutions across MENA, and we deploy open-source AI inside their perimeter. Same conviction, two expressions. The data stays where the bank does.

DelText AI · product overview 02:14
26
Institutions · MENA + India
8
Countries · gcc · levant · egypt · india
24×7
Support · 15-min response
0
Bytes · leaving the bank

The institutions that already run on us.

Twenty-six institutions across the GCC, Egypt, and India have engaged DeltaSoft for open-source database operations and AI tooling — including the largest bank in the region. A selected list, not exhaustive.

Qatar National Bank
Qatar
RAKBank
United Arab Emirates
Emirates Development Bank
United Arab Emirates
Sharjah Islamic Bank
United Arab Emirates
United Arab Bank
United Arab Emirates
Bank of Sharjah
United Arab Emirates
Ajman Bank
United Arab Emirates
Al Masraf
United Arab Emirates
National Bank of Fujairah
United Arab Emirates
National Bank of Umm Al Quwain
United Arab Emirates
QNB Al Ahli
Egypt
ADCB Egypt
Egypt
Arab National Bank
Saudi Arabia
Alinma Bank
Saudi Arabia
Gulf Bank
Kuwait
Burgan Bank
Kuwait
Gulf International Bank
Bahrain
Al Salam Bank
Bahrain
Bank ABC
Bahrain
Ahli United Bank
Bahrain
National Bank of Oman
Oman
Bank Dhofar
Oman
Masraf Al Rayan
Qatar
Ministry of External Affairs
India
DP World
United Arab Emirates · corporate
Monochrome Event Mgmt
United Arab Emirates · corporate
The list spans retail, corporate, Islamic, and development banking mandates — plus selected sovereign and corporate engagements.
26 institutions · selective list

Open-source databases, middleware, and infrastructure — run to a bank-grade SLA.

Twenty-four hours a day, in every country we serve. Our PostgreSQL, MySQL, MariaDB, MongoDB, Redis, Tomcat, and Kubernetes practice runs on a published service contract — not best-effort. Same engineers run the live banks, every hour of every day.

24×7

Always-on support

Day, night, weekend, holiday. Dedicated team operates around the clock from a globally distributed pool.

15min

Initial response

Every ticket acknowledged inside 15 minutes via Zoho Desk. No queue, no triage delay.

2days

Root-cause analysis

For every incident, we deliver a written RCA and an implementation plan inside 48 hours.

2wks

Full resolution

Target window from incident open to permanent fix, across patching, tuning, replication, and recovery.

Global resource pool

Engineers across regions and time zones. Continuity isn't dependent on one office, one country, or one person.

One partner for the whole stack.

Most banks pay three to five vendors for fragmented open-source coverage — a Postgres specialist here, a Kubernetes shop there, a separate Tomcat contract, a different consultancy for the AI work. We are one of them. Unified SLA, unified ownership, one team that knows where the database ends and the application begins.

Layer 01

Database & Storage

PostgreSQL MySQL MariaDB MongoDB Redis pgvector
Patroni clusters, pgBackRest, pgBouncer, InnoDB Cluster — HA, DR, PITR, and replication patterns proven across 26 institutions.
Layer 02

Middleware & Infrastructure

Kubernetes AKS · EKS CloudNativePG Operator Tomcat nginx Prometheus Grafana
Declarative database automation, container orchestration, observability and alerting tailored to each bank's environment.
Layer 03

AI & Data

Llama 3 Mistral Python · FastAPI Full-text search Vector search
Open-source LLMs deployed inside the bank's perimeter. Structured and unstructured data brought into one credit-grade workflow.

Mission-critical work, already running.

Four engagements that show the shape of what we do — from sovereign passport infrastructure to retail-banking mobile backends to a sixty-microservice migration off Oracle. Named clients on request, under NDA.

01
Oracle → PostgreSQL migration · logistics

50+ microservices migrated off Oracle — zero downtime.

Client
Global port-management leader in smart end-to-end supply chain logistics.
Challenge
Migrate 50+ Oracle microservices to PostgreSQL for scalability and cost.
Solution
Phased migration strategy with zero downtime and full data-integrity validation at each cutover.
Significant reduction in database-related operating cost.
All microservices in production on PostgreSQL.
02
High-availability PostgreSQL · retail banking

99.999% uptime for a regional bank's mobile-banking database.

Client
A leading international bank in the MENA region.
Challenge
Highly available PostgreSQL for a critical mobile app serving retail customers.
Solution
Patroni-based PostgreSQL cluster with automatic failover and continuous replication.
99.999% measured uptime on the mobile-banking database.
Sub-second failover validated under load.
03
PKI database · ministry of external affairs

100% uptime since deployment for an e-passport PKI handling millions of verifications daily.

Client
Ministry of External Affairs, India.
Challenge
Maintain a super-critical 24/7 PKI database for the e-passport system — the workload behind airport operations worldwide.
Solution
Hardened PostgreSQL database with real-time replication and multi-site redundancy.
100% uptime since deployment.
Millions of passport verifications processed daily across global airports.
04
CreditLens support · 15+ MENA banks

40% performance lift on Moody's CreditLens across 15+ banks; 100% recovery in every DR scenario.

Client
Over 15 banks across the MENA region running Moody's CreditLens for corporate credit.
Challenge
End-to-end support: high availability, backup, point-in-time recovery, and 24/7 operations.
Solution
Bank-specific PostgreSQL stacks — HA clusters, PITR, and round-the-clock support per institution.
~99.5% reduction in system downtime, average across clients.
~40% performance improvement via database optimization.
100% successful recovery across all DR exercises.
05
Cloud-native AI on Kubernetes · credit risk

AI-powered credit risk analysis running on AKS with PostgreSQL + pgvector — auto-scaling, declarative, open-source end to end.

Kubernetes (AKS) CloudNativePG Operator PostgreSQL pgvector Full-text search
Challenge
Run an AI-driven credit-risk workload that ingests structured *and* unstructured data, scales with demand, and stays inside the bank's perimeter.
Solution
Declarative PostgreSQL on AKS via CloudNativePG Operator. pgvector for embeddings, full-text search for documents. Automatic infrastructure scaling with the workload.
Outcome
Same database engine handling transactional, search, and vector workloads. No data leaves the AKS perimeter. Scaling on infrastructure, not on a vendor's pricing schedule.

Banks should own their data and their AI.

Open source is no longer the cheap-but-risky alternative. It is the only stack where a bank can verify what runs, where it runs, and what it sees. The same logic that argues for open-source databases now argues for open-source AI. We build for both.

01

Data stays inside the perimeter.

No statement, customer record, or policy clause leaves the bank's network. Open-source LLMs run on the bank's own hardware. No vendor API egress.

on-prem · air-gapped · uae / gcc residency
02

The stack the bank can leave.

PostgreSQL, MySQL, MariaDB, MongoDB, Redis, Tomcat, Kubernetes, Llama, Mistral — the bank can switch vendors, support partners, or model versions without rewriting its operations. The contract earns the renewal.

portable · vendor-neutral · standards-based
03

Auditable, not a black box.

Every database transaction, every model prompt, every retrieved document is traceable. Risk, Compliance, and Internal Audit get the evidence they need without a vendor disclosure request.

traceable · reproducible · audit-ready

Credit memos drafted by AI, judged by your analyst.

DelText ingests audited financials, management accounts, statements, and policy clauses — structured and unstructured — and produces a structured first-draft credit memo in the bank's own format. The analyst keeps the judgment. The bank keeps the data. The credit decision is unchanged.

DelText

Open-source LLMs, deployed inside the bank.

Open-source models. Bank's own hardware. No data leaves the network — including for some of the region's largest institutions. Every claim, citation, and policy reference is traceable to the source document.

Llama 3 · Mistral on-prem 43 sections traceable citations
deltext.ai · corporate · peer regimes · building materials
DelText product screenshot — MENA building materials credit-regime quadrant
// Indicative output · MENA building-materials peer credit-regime quadrant
Certified partner
Moody's Analytics
CREDITLENS · MENA REGION

DeltaSoft is a certified partner of Moody's Analytics for PostgreSQL, supporting CreditLens deployments at 140+ banks across the Middle East. CreditLens runs on PostgreSQL; we are the regional support partner for that stack. The certification came on the back of our structured support service — the same SLA you saw above.

The same team that runs the live banks — also answers your call.

Engineering, sales, and operations across Dubai and Larnaca. UK-registered. Small, named, and reachable.

Talk to the team that runs your stack.

New engagement, existing-client support, or a walk-through of what DelText might look like inside your bank — three paths, all answered fast.

For new conversations

Book a call.

Thirty minutes on the team's calendar. We come prepared for your bank's profile — obligor count, current stack, where you'd want a pilot.

Open the calendar
For existing clients

Support tickets.

Issue with a live PostgreSQL, MySQL, MariaDB, or DelText deployment? Initial response inside 15 minutes via Zoho Desk.

Open a ticket