01
Legacy core
One change, one release window, one big risk - the vendor owns the roadmap, the venue owns the risk.
Cloud-native market infrastructure
One core for trading, clearing, settlement and data
UNIFIN is a production platform built as microservices: 49 services, 9 client portals and 43 databases that launch new market products by configuration - and go live as a pilot in 2-4 weeks.
0microservices in production
0client and staff portals
0documented REST endpoints
99.9%money-path availability SLO
The case
Why the industry needs a new architecture
Legacy monoliths limit venues by release windows, integration cost and vendor roadmaps. The market needs infrastructure that scales with volume and absorbs new asset classes.
02
Time to market
A new product takes 12-18 months of vendor backlog. Clients and regulators do not wait that long.
03
Growing volumes
Turnover and participants grow every year while the old core scales only by buying more hardware.
04
Integration and interoperability
Every bank, CSD and regulator needs a connector; point-to-point links multiply and fail silently.
05
New asset classes
Digital assets, commodities and derivatives arrive faster than a monolith can absorb them.
06
Cost of ownership
Per-seat and per-module licences make growth more expensive than the revenue it brings.
Our answer - one platform
Already built, already running. Base licence for the core plus modular add-ons for microservices and data products: you pay for what you switch on.
Microservices, not a monolithUpgrade one service, not everything.
Launch by configurationNew products are settings, not projects.
Scale on KubernetesAdd nodes from pilot to national scale.
Standards instead of point-to-pointFIX, ISO 20022, REST, SFTP, webhooks.
New assets on the same coreClearing, settlement and audit are reused.
Predictable cost of ownershipFewer systems, one team, one run cost.
Platform
Cloud-native by construction
Every function is an independent service with its own database, release cycle and contract - the same code runs a 3-node pilot and a national deployment.
Independent services
49 microservices, 43 per-service databases, transactional outbox with schema-versioned events and DLQ replay.
One core, many products
Trading, post-trade, payments, risk, depository and data reuse the same identity, audit and reference data.
Operated, not just delivered
Kubernetes and Helm, rolling updates with zero downtime, Prometheus/Grafana dashboards and burn-rate SLO alerts.
cloud-nativeKubernetes + Helmevent-driven (outbox)
43 own databasesindependent releasesAPI-first
Company
A team that runs market infrastructure, not slide decks
UNIFIN builds and operates exchange-grade systems. The platform on this page is in production: 49 microservices, 9 portals, 43 databases and 978 documented endpoints, delivered and supported by teams in Singapore, Abu Dhabi and Kazakhstan.
UNIFIN is an engineering company for market infrastructure. We design, build and operate exchange-grade platforms - trading, post-trade, depository, payments, risk and data - for venues, banks and financial groups, with teams in Singapore, Abu Dhabi and Kazakhstan.
The platform on this page is in production, not a concept. It replaced monolithic risk with 49 independent services, moved releases from yearly windows to weekly waves, and turned new products into configuration instead of multi-month projects.
We stay for the operational lifetime: pilot, rollout, service levels, disaster drills and support - one accountable team from the first scope call to national scale.
Where our experience sits
Venues and exchangesmatching and auctions, market data, indices, listings and supervision - the full venue stack.
Post-trade and depositoriesclearing and netting, DVP settlement, custody accounts and corporate actions with a complete audit trail.
Banks, payments and FXISO 20022 and MT rails, bank file exchange, reconciliation, FX conversion and hedging instruments.
Regulators and complianceregulatory reporting, sanctions screening, market surveillance and the evidence to prove it.
Capabilities
What the platform covers out of the box
Twelve core capability groups are already running; each one is a microservice with a published API and its own release cycle.
Directions
Six directions of the product
Trading and market operations
Order book, auctions, OTC desk and RFQ, primary placements, market data and indices.
Clearing, settlement and CSD
Clearing cycle, netting, margin calls, DVP settlement, custody accounts and corporate actions.
Payments, banking and FX
Payment orders, bank files and API channels, ISO 20022 and MT messages, FX conversion.
Risk, compliance and supervision
Pre-trade limits, sanctions screening, surveillance cases and regulatory reporting.
Client services and documents
Onboarding and KYC, nine portals, document exchange, signatures and notifications.
New markets and growth
Derivatives, digital assets, commodities, AI analytics, data subscriptions and training.
Potential data products
Products you can launch on the same core
Each product reuses the platform rails - clearing, settlement, depository, identity and audit - and adds its own store, UI and commercial model.
Market data hub
Data locked inside the core; history requests take days.
Live streams, history and files with entitlements and metering.
Details
Corporate actions portal
Events run on spreadsheets; deadlines are missed.
Announce, approve, execute and pay with a full audit trail.
Details
OTC commodity trading
Deals closed in chat; delivery and payment decoupled.
RFQ, deal tickets, clearing and DVP settlement on one rail.
Details
Bond platform
Allocations in Excel; coupons processed by hand.
Placement book, secondary trading, coupon and redemption runs.
Details
Index products
No independent index; licensing uncontrolled.
Published methodology, streamed values, licensing and metering.
Details
AI analytics
Reports arrive late; data is fragmented.
Natural-language questions over live data; dashboards and copilot.
Details
Digital assets platform
Tokens cannot be admitted safely to the market.
Registry with lifecycle, admission rules and screening; classic settlement.
Details
Derivative trading
Margin calculated late in spreadsheets.
Futures and options with SPAN and variation margin, cash settlement.
Details
AI agentic layer
Operations drown in routine; knowledge stays in heads.
Agents draft and summarise; people approve; every step is audited.
Details
Click a product - the card opens with what it is, why it pays off with us and which pains from the top of the page it closes.
Launch map
What you can launch - by industry and by client type
The same core assembles into a different product set every time. Pick who you are - the list updates instantly; the full scope is agreed on the call.
Full catalog
Everything you can build on this platform
One hundred and ten product ideas across four tiers - from the shortlist we would launch first to the long tail that makes sense on partnerships. Every row opens a card: audience, platform basis, why we win and the main risk. Commercial terms are agreed on the call.
API surface
The API, mapped to the domain model
978 REST endpoints behind 126 gateway routes, grouped the DDD way: layers, bounded contexts and the endpoints that exist in the code today. Pick a layer, then a domain - the panel shows representative routes and how they are exposed.
Inventory snapshot from the codebase: controllers grouped by service. Full OpenAPI contracts are shared under NDA - the section above lists the platform-wide counts.
Technology
Standard, open and containerised
No exotic dependencies: a team familiar with Java, PostgreSQL and Kubernetes can operate the platform from day one.
RuntimeJava 21 (LTS), Gradle 8.9
BackendSpring Boot 3.5, Spring Cloud 2025, Flyway
PortalsNext.js 16, React 19, TypeScript 5
DataPostgreSQL 17 per service (43), TimescaleDB, Redis 7 + Sentinel
MessagingRabbitMQ 3.13 quorum queues, transactional outbox, DLQ
IdentityKeycloak 26.9 (OIDC, JWT), API keys and quotas
PlatformKubernetes + Helm, Docker Compose for pilots
ObservabilityOpenTelemetry, Jaeger, Prometheus + Grafana, Loki
Security & QAVault, Trivy, OWASP ZAP, 88 guard scripts, Testcontainers
CI/CDGitLab pipelines with 74 automated checks per merge
Minimal pilot cluster
3 nodes x 8 vCPU / 32 GB / 500 GB NVMe plus PostgreSQL primary and replica. Runs the full platform with production settings.
HA production start
6+ nodes x 16 vCPU / 64 GB, Patroni x3, RabbitMQ x3, Redis Sentinel x3. Money-path pods spread with anti-affinity, rolling updates with zero downtime.
Growth without redesign
Add nodes horizontally: the same Helm chart deploys 3 or 30 nodes; databases stay small and owned.
Scale
Performance you can hold
Same code from pilot to national scale - capacity grows by adding nodes, not by rewriting the platform.
0accounts on one installation
0orders per day at peak profile
0events per day through the backbone
0ticks per second
Service levels we hold
Money-path availability (28 days)99.9%
Money-path latency, p95< 100 ms
Error rate, Tier 1< 0.1%
Core platform (Tier 2), p95< 500 ms
Chaos scenarios replayed monthly8
Trading core, in numbers
Order-to-acknowledge, p95< 100 ms
Matching enginein-memory, price-time priority, event-sourced
Engine shardinginstrument groups split across shards - add a shard, not a rebuild
Instruments at oncetens of thousands per cluster, thousands per shard
Market data fan-out12,000+ ticks per second to portals and partners
Latency budget: matching and validation stay in memory; storage, reporting and AI work asynchronously on events - the hot path never waits for the disk.
How microservices scale
All 49 services own their database and scale independently: matching needs CPU, the data hub needs storage, portals need replicas. Under load we add pods for the hot service only - the rest keeps running, no maintenance window.
- Horizontal pods (HPA) per service with money-path anti-affinity across nodes.
- Reads scale out through read models and the event backbone, not by locking tables.
- Rolling updates with zero downtime; the platform grows from 3 nodes to 12+ without code changes.
- Sharded matching and market data keep latency flat as instruments and volumes grow.
Integration
Connectivity that partners recognise
Standard channels on the edge, uniform rules inside: traceId end to end, idempotency keys, outbox in the same transaction and fail-closed money handling.
Inbound
- FIX 4.2 / 4.4 / 5.0 gateway with drop copy
- Bank REST with mTLS and SFTP file exchange
- B2B participant gateway with API keys and quotas
- Market data feeds, sanctions lists, batch uploads
Inside the platform
- API gateway with 126 routed prefixes and canary weights
- Keycloak identity, roles and service accounts
- Event backbone with schema versioning and DLQ replay
- ISO 20022 and MT message builders in the integration hub
Outbound
- Nine portals plus WebSocket / SSE live streams
- SWIFT / RTGS gateways, NACHA and ISO 20022 messages
- Reporting distribution by e-mail, SFTP and object storage
- Webhooks and nightly extracts for partner systems
traceId end to endidempotency keysoutbox in the same transactionDLQ with replayfail-closed money pathsandbox from day one
Build with us
Launch your product on 49 microservices
Pick the bricks your business needs. Every brick is a microservice with its own database and release cycle - you compose a product instead of writing a platform from scratch.
API gatewayIdentity & accessConnectors inConnectors out
Event backboneNotificationsAudit & trailLogging & metrics
Reference dataData storesTime seriesFile store
Report engineRules & limitsScheduler & jobsSecrets & vault
AI analytics & dashboards
Market and operational analytics with an AI copilot.
Market surveillance
Abuse detection, alerts and investigation cases.
OTC bond trading
RFQ, deal tickets, clearing and DVP settlement.
IPO platform
Placement book, allocations and listing day.
Corporate reporting
Issuer reports, corporate actions and signatures.
Document exchange
Secure document flow with a full audit trail.
Or your own product: gateway, identity, event backbone, rules and reports become a working pilot cluster with your scenario, users and data.
Next step
Start directly with a pilot
No long pre-sales: choose a scope and we deploy a working pilot cluster with your instruments, users and data - typically in 2-4 weeks.
About the company
Engineering and operating market infrastructure for exchanges and financial groups, with teams in Singapore, Abu Dhabi and Kazakhstan.
info@unifin.tech - corporate enquiries, partnerships
More about the product
Platform capabilities, data products and AI services: case studies, screen tours and the commercial model.
uni@unifinai.com - demos, pilots, commercial questions
Thank you - your request is on its way
We reply within one business day. If it is urgent, write to uni@unifinai.com.