Deterministic risk controls at the execution layer

Cloud-Native Quantitative Infrastructure

Aesir Core builds the execution backbone for systematic trading — pairing elastic, containerized compute with algorithmic risk-management engines that enforce portfolio limits in-line, before an order ever reaches a venue.

View Documentation Request Access

Private beta · Onboarding systematic funds and proprietary trading desks

aesir-risk-engine — pre-trade check

$ aesir risk simulate --strategy momentum-eu --notional 4.2M

[ingest]   ws://md.aesircore.internal   stream established · 14 venues
[compute]  k8s/exec-pool-3              6 pods · autoscale enabled
[limits]   gross_exposure   4.20M / 5.00M   PASS
[limits]   single_name_max  1.8% / 3.0%     PASS
[limits]   intraday_dd      0.4% / 2.0%     PASS
[limits]   venue_conc       38%  / 35%      BREACH

[halt]     order rejected pre-submission · policy venue_conc
→ audit record written · immutable log

Architecture

Built for the failure modes that matter

Three layers, each independently observable and independently recoverable — so a fault in one never becomes an uncontrolled position in another.

High-Frequency Data Ingestion

Persistent WebSocket connections to venue and vendor feeds, normalised into a single tick schema on arrival. Sequence-gap detection, automatic re-subscription, and replay from the durable log after any disconnect.

· WebSocket / FIX gateway
· Order-book reconstruction
· Gap detection & replay

Containerized Execution

Every strategy ships as an immutable Docker image and runs as an isolated workload on Kubernetes. Horizontal autoscaling for backtest sweeps, pinned resources for live execution, identical binaries across both.

· Kubernetes orchestration
· Immutable image promotion
· Blue/green strategy rollout

Hard-Coded Risk Limits

Automated portfolio protection enforced in the order path, not in strategy code. Gross and net exposure, single-name concentration, and intraday drawdown ceilings are evaluated pre-submission — a breach halts the order and writes an immutable audit record.

· Pre-trade limit checks
· Automated kill-switch
· Immutable audit trail

Security & Infrastructure

Isolated by default. Trusted by nothing.

The engine runs in secure, isolated cloud environments with no shared tenancy between clients. Every service-to-service call is authenticated and authorised independently — a zero-trust posture in which network position grants no privilege on its own.

Per-tenant isolation: Dedicated namespaces, dedicated credentials, and separate encryption keys per client. No strategy state or market data crosses a tenant boundary.
Zero-trust service mesh: Mutual TLS between all internal services, short-lived workload identities, and least-privilege IAM scoped per component rather than per environment.
Encrypted and evidenced: Encryption in transit and at rest throughout, with append-only audit logging across the order path. SOC 2 Type II readiness work is in progress.

Deployment posture

Infrastructure is defined as code and reproducible per region, so a client environment can be rebuilt from a clean account without manual steps.

Design target

99.99%

Multi-AZ control plane

Risk check

Pre-trade

In the order path

Tenancy

Isolated

Namespace per client

Compliance

In progress

SOC 2 Type II readiness

Kubernetes Docker Terraform mTLS OpenTelemetry

Contact

Request platform access

Aesir Core is onboarding a limited number of systematic funds and proprietary desks during private beta. Tell us about your stack and we'll follow up with architecture documentation.