Market Infrastructure & Exchanges
Market data in, orders out, books reconciled.
- Low-latency cross-venue execution
- Backtests with sim/live parity
- Matching, clearing, reporting
- Risk limits, KYC/AML
order book, asks / bids
We build and run demanding systems.
Market infrastructure, blockchain and applied AI.
From first design to production.
latency-monitor · exec-core / tick→orderprod · ld4-rack07
Tick-to-order histogramlog/log · bin width 2.5%
Percentiles
p99 budget 20.0 µs · ● met, 60 of 60 min
By venue5 venues · order gateway → wire
p50 / p99 per secondp50 p99 last 90 s
book-viewer · BTC-USD · V2seq
Laddersize BTC · cum.
Depthcum. BTC · ±10 levels
Time & sales
chain-watch · solana mainnet-betaprogram 7xKq…p9Tz
recon-assist · break triageclassifier v3.2 · claude-sonnet-4-5
Triage trace · 7 spans · 1,236 ms
Model qualityops-labelled · last 7 days
Break queue
service-health · prod / eu-centralk8s 1.33 · 3 node pools
Kafka consumer lagmsgs · live
Deploysrolling, canary 10%
| time (UTC) | service | version | canary | duration | result |
|---|---|---|---|---|---|
| 09-23 09:14 | orders-api | v4.18.2 | p95 +0.4 ms | 6m 12s | promoted |
| 09-22 16:40 | md-ingest | v2.9.0 | err 0.00% | 8m 03s | promoted |
| 09-21 11:02 | notifier | v1.31.1 | err 1.9% | 3m 41s | rolled back |
| 09-21 11:20 | notifier | v1.31.2 | err 0.01% | 4m 02s | promoted |
| 09-19 14:27 | risk-engine | v7.2.5 | p95 −0.2 ms | 7m 55s | promoted |
| 09-18 10:05 | api-gateway | v3.4.0 | p95 +1.1 ms | 9m 30s | promoted |
| 09-17 15:48 | ledger-worker | v5.0.3 | err 0.00% | 5m 17s | promoted |
| 09-16 09:31 | orders-api | v4.18.1 | p95 −0.3 ms | 6m 40s | promoted |
| 09-15 13:12 | md-ingest | v2.8.4 | err 0.00% | 7m 58s | promoted |
| 09-12 17:05 | risk-engine | v7.2.4 | p95 +0.1 ms | 7m 21s | promoted |
| 09-11 10:44 | api-gateway | v3.3.9 | err 0.00% | 8m 49s | promoted |
Representative views of production tooling. Data anonymized.
10+
years in production systems
100+
client projects delivered
50+
clients worldwide
99.9%
system uptime
One team,
end to end.
end to end.
What we build
Blockchain & Digital Assets
On-chain programs and the services around them.
- Solana programs in Rust/Anchor
- EVM contracts in Solidity
- DeFi, tokens, NFTs, DAOs
- Wallets: Fireblocks, Privy, Turnkey
blocks, one after another
Applied AI for Financial Systems
ML, automation and LLMs on top of the financial core we build.
- ML on market data, ClickHouse
- GPU training and fine-tuning
- Automated reconciliation and reporting
- LLM systems in client apps
data → model → app
Product & Data Platforms
Full-stack products and the infra under them.
- Web, mobile and desktop apps
- APIs, microservices, SSO
- Petabyte data, sub-second queries
- Kubernetes, Terraform, Grafana
services and how they connect
Work
venue-quotes · BTC-USD · top of book UTC
Best bidacross 5 venues
Best askacross 5 venues
Cross-venue spreadUSD · 5 / 5 feeds up
feeds normalized to one clock · quotes timestamped at the NIC
Multi-venue execution
One execution engine connected to 5 venues, with risk limits on every order. Built for clients and run in-house.
Our in-house system executes across 5 venues at internal tick-to-order under 20 µs.
backtest-runner · jobskdb+ tick data · build a1f3c9e
| job | strategy · period | progress | percent | status |
|---|---|---|---|---|
| bt-5521 | mean-rev v14 · 2019–2026 | 62.4% | running | |
| bt-5522 | mean-rev v15 · 2019–2026 | 18.1% | running | |
| bt-5520 | mean-rev v14 · 2026-09-22 | 100.0% | parity 99.8% | |
| bt-5519 | basis v3 · 2021–2026 | 100.0% | done |
Events replayed4,118,402,917today, both running jobs
Replay speed1,640×vs wall clock
Code pathsharedsame build as live
Backtesting with sim/live parity
kdb+ tick-data ingestion and a backtester that shares its code with the live execution path.
The same strategy code runs in simulation and live.
match-engine · partitionsprimary ld4 · standby fr2 in sync
| partition | instruments | orders / s | trades / s | match p99 µs | sequence |
|---|---|---|---|---|---|
| P1 · perps | 38 | 4,210 | 612 | 7.9 | 921,806,433 |
| P2 · spot | 196 | 2,380 | 341 | 6.4 | 604,118,552 |
| P3 · futures | 104 | 1,150 | 97 | 5.8 | 311,907,004 |
| P4 · options | 74 | 890 | 12 | 9.2 | 128,550,630 |
Instruments4124 partitions
Trades today1,184,630all partitions
Replication lag0 msgsstandby in sync
Exchange systems, matching to reporting
Matching engine, order management, clearing, reconciliation and regulatory reporting.
99.9% uptime on integrations.
Read the case study: Exchange systems, matching to reporting →
ingest · kafka → clickhousemd_cluster · 6 shards × 2 replicas
| table | rows / s | total rows | active parts | kafka lag |
|---|---|---|---|---|
| md.quotes | 184,200 | 9,441,870,112 | 38 | 1,204 |
| md.trades | 41,300 | 1,952,306,418 | 21 | 188 |
| md.book_l2 | 96,800 | 4,870,155,902 | 44 | 610 |
| oms.orders | 2,150 | 108,442,917 | 9 | 0 |
Insert batches1 / sper table, async inserts
Compression11.8×ZSTD + Delta codecs
Replicas12 / 12healthy
ML research on market data (ClickHouse)
A ClickHouse analytics store feeding ML research on large-scale market data.
Billions of rows of tick data, queried in under a second.
Read the case study: ML research on market data (ClickHouse) →
Representative views of production tooling. Data anonymized.
Next step
Have a hard system to build?
Tell us what it has to do. We will tell you how we would build it.