Ashva
liveMulti-tenant tote and fixed-odds betting platform, in production.
Product engineering · Data & AI · 24×7 operations
We build and run our own platforms.
We build yours to the same standard.
Protocolzone develops custom software platforms, integrates business systems, and implements AI and automation into operational workflows. We work with your team from planning and engineering through launch and ongoing operation.
Bring a business goal, a platform idea or an existing system. A technical specification is optional.
Our products
Products we build and operate.
Ashva, AnkEDGE and AmshPOS put our engineering into daily operation. The same experience goes into the platforms we build for clients.
Services
Three ways we engage — usually all three.
Most engagements start with a data problem, a platform build, or an existing system that needs to make smarter decisions — and end with us running the result. That continuity is the point: the team that built it is the team on call.
Platform to Platform
Custom platform development & modernisation
Build a new platform, connect it to your business, or modernise the system you already run. Customer applications, operational consoles, APIs and automation, with a team that stays through launch and operation.
- Custom platforms
- API & integration
- Workflow automation
- AI modernisation
Data to Data
Data engineering & AI implementation
Turn disconnected feeds into usable data and put models into the workflow: forecasting, computer vision, text classification and decision support, with monitoring and a path for human review.
- Data pipelines
- Forecasting & models
- Computer vision & NLP
- Data governance
24×7 Operations Desk
Monitoring, incident response & release support
Our engineers operate production systems across cloud-native, hybrid and legacy estates. Agreed shift coverage, runbooks and escalation paths keep responsibility clear when an incident occurs.
- 24×7 monitoring
- Incident response
- Release management
- Continuous improvement
AI, automation and modernisation
The system keeps running. The decisions get better.
Forecasting prices, identifying animals from video, and classifying civic feedback are different problems. Our delivered work connects models to the data, interfaces and people who act on their output.
Forecasting for a perishable-goods manufacturer
Recipe costs, sales and wastage brought into pricing and optimisation across more than 100 SKUs. A 2021–22 engagement connecting models to commercial decisions.
Read the implementation →Computer vision with operator feedback
Individual tigers identified from stripe patterns in live video. The 2018 implementation linked sightings to a map and used operator corrections to improve the model.
Read the implementation →Automation inside a live platform
A deployment failure detected from its logs, a remediation prepared, and an engineer confirming the change. Automation removes repetitive steps while keeping consequential decisions visible.
Read the implementation →Your technology partner
One team from the first release to the next.
Our own platforms keep changing after launch. We bring the same continuity to client work: architecture, implementation, integrations and the operational knowledge needed to improve the result.
01
Define the next release
For a new platform, agree the users, workflows, interfaces and first useful release. For an existing system, review the constraints and plan changes around the business already using it.
02
Build and integrate
Make the platform work with its data sources and partner systems. Review working software, test failure paths and prepare a repeatable release process.
03
Operate and improve
Agree monitoring, escalation and support before launch. Use operational feedback to prioritise improvements, with the build team available to work on the underlying system.
Engineering in practice
The details change with the domain.
Explore the routing, data and operational constraints behind the platforms we build, and the applications of that experience.
Tote / pari-mutuel
Every bet is
a routing decision.
Bypass it to a tote host and you earn commission with nothing at risk. Retain it and you keep the full margin and the full liability. That choice has to be made per bet, in the moment, against a stake limit and a liability position that are both already moving.
We built and operate a multi-tenant tote and fixed-odds betting platform. A rules engine routes each bet to an external tote host, to a fixed-odds operator, into in-house risk, or splits it — across win, place, exotic and multi-leg pools, with scratchings propagating mid-market.
Bet → routing rules → bypass, retain, or share the risk
Wagering at peak
Melbourne Cup day is
the load test.
Peak load is one Tuesday in November and a run of Saturday metro meetings — not an even quarterly curve. Capacity planned against an average is capacity that fails on the day that counts.
One multi-tenant platform carrying every brand through the same peak, with per-tenant live money-in, exposure and risk position visible in a single console rather than reconciled overnight.
Feeds → routing, pricing, settlement → channels
Mining & resources
The pit does not stop
for your batch window.
Haul fleets, crushers and conveyors emit continuously. If the telemetry lands in an overnight batch, every anomaly is discovered after the shift it happened in.
Continuous ingestion of high-frequency sensor streams, with drift and anomaly detection running against the live signal rather than yesterday’s extract.
Vibration signal → threshold band → anomaly flagged in-shift
Insurance
Claims arrive in clusters.
Capacity should not be flat.
A hail event does not spread its claims evenly across the month. Provisioning for the mean means queueing for days exactly when responsiveness is most visible.
Elastic intake sized against burst arrival rather than average, with fraud and leakage cohorts surfaced at triage instead of in a retrospective audit.
Burst arrivals → triage queue → cohorts separated
Trading
Latency is not a metric.
It is a position.
Every millisecond between a tick and your decision is inventory you are holding at a price that has already moved. Slow is not a performance issue, it is an exposure.
Low-latency market data paths, deterministic order handling, and risk computed on the live book rather than on a periodic snapshot.
Tick stream → book state → risk on the live position
Exchange & market operations
Matching is deterministic,
or it is a dispute.
Price-time priority has to hold under every replay, every failover and every audit. A matching engine that is only usually fair is a legal problem waiting for a busy day.
Deterministic matching with strict price-time priority, replayable event logs, and settlement reconciled against the same ordered stream the match came from.
Bid / ask queues → price-time match → replayable log
Digital assets
A reorg is a data problem
before it is a finance one.
Chain state is provisional until it is not. Treating a recent block as final is how a balance becomes wrong quietly, in a system that reports itself as healthy.
Confirmation-aware ingestion that treats depth as a first-class field, reverses cleanly on reorg, and never reports provisional state as settled.
Blocks append → fork detected → state reversed to common ancestor
Government & civic
Records outlive
the systems holding them.
A citizen record has a longer life than any platform it has lived on. Every migration that loses provenance makes the next one harder and the audit weaker.
Migrations that carry lineage with the data, so provenance survives the move and the next platform inherits a defensible history rather than a flat export.
Legacy stores → lineage preserved → consolidated register
Retail & FMCG
The till still works
when the link drops.
Connectivity across a store estate is not uniform and never will be. A point of sale that needs the network to take a sale is a point of sale that closes the lane.
Offline-first transaction capture with a durable local queue, deterministic reconciliation on reconnect, and no duplicate settlement when the link returns.
Offline queue → reconnect → reconciled once, not twice
24×7 Operations Desk
Your quiet hours are
someone else’s peak.
Incidents do not schedule themselves against your roster. The expensive ones land when the fewest people are awake and the least context is available.
Managed operations across cloud-native, hybrid and legacy estates — run by the engineers who wrote the code, 24×7 across time zones, with Australian racing hours covered as a matter of course.
Coverage window → incident → escalation path
What we work on
- Tote / pari-mutuel
- Per-bet routing
- Fixed-odds pricing
- Same-race multi
- Algorithmic trading
- Strategy backtesting
- Market data processing
- Exposure & position analytics
- ML in production
- Predictive modelling
- Computer vision
- Event streaming
- Feed normalisation
- Smart POS
- Order tracking & inventory
- Waste & markdown optimisation
- Multi-tenant tenancy
- Real-time risk
- ADW wallets
- Cloud cost optimisation
- Legacy migration
Industries
Where we've actually delivered.
7 areas have delivered work behind them. 6 describe capabilities we can apply, with no delivered reference claimed.
Wagering & Racing Fixed odds, tote, exotics Delivered Trading Market data, risk, settlement Delivered Retail & FMCG POS, inventory, loyalty Delivered Gambling & iGaming Wallets, content, RG enforcement Delivered 24×7 Operations Monitoring, incident response Delivered Technical Support Escalation, release management Delivered Government & Civic Video analytics, sentiment, registers Delivered Mining & Resources Sensor telemetry, condition monitoring Capability Insurance Claims intake, fraud cohorts Capability Exchange Matching, price-time priority Capability Digital Assets Confirmation depth, reorg safety Capability Healthcare & Diagnostics Records, imaging pipelines Capability Manufacturing & Supply Chain Telemetry, traceability Capability
Why operators use us
Coverage that lines up with your peak, wherever it is.
We serve clients globally, 24×7, from Ahmedabad. The sharpest example is Australian racing: our operations coverage includes Saturday metro meetings and the spring carnival, when an incident costs the most and local teams are hardest to staff.
We don't have an Australian office. What we have is delivery history in the market, familiarity with the regulatory shape of it, and engineers on shift when your traffic peaks. Platform development and AI implementation for Australian teams →
-
The build team runs it
No handover to a separate managed-services vendor who has never seen the code. The people who wrote the settlement logic are the people paged when it misbehaves.
-
Tenancy designed around your platform
Operating Ashva taught us where tenant isolation has to hold: data access, background jobs, configuration and reporting. We plan those boundaries with you before building.
-
Regulatory work treated as engineering
Certification and jurisdictional rules are requirements with test cases, not paperwork bolted on before launch.
Tell us what you want to build or improve.
A new platform, an integration, a workflow to automate or a system that needs better decisions. Tell us the goal and the constraints you know. We will work through the next step with you.