Co-founder · Lead architect

GoToCrypto

2024–present

A trading platform that can explain every decision.

Product screens

01 / 05

GoToCrypto execution dashboard showing open positions, balances, exposure and combined P&L — Execution dashboard: open positions, balances and combined P&L. Dev environment, synthetic market feed.

Overview

I co-founded GoToCrypto and built its event-driven crypto trading platform: every decision is recorded, every action is reconciled against the exchange, and the system recovers safely from restarts.

Problem statement

Systematic crypto trading has to be trustworthy, not just easy to demonstrate. The platform must recover from restarts, replay events safely, keep market data free of gaps and show, for any position, exactly why it exists.

Solution

Decisions as durable records.

I record every signal as a candidate before anything touches an exchange. Its identifier doubles as the idempotency key, and skipped signals stay on record, so the audit trail covers what the system chose not to do.

A single exchange boundary.

I route all exchange calls through one gateway and reconcile against source events to establish what actually filled. A successful request is never treated as a confirmed fill.

Gap-free market data.

I ingest trades, candles, funding and open interest from Hyperliquid, with Binance as the historical source. The data is de-duplicated, back-filled from a central archive, stored in Parquet and replayable. Gaps are detected and healed, not hidden.

Safe signing and operations.

I sign orders with a dedicated agent wallet that the exchange restricts to trading actions, so no long-running service holds a main wallet key. Kill switches and safety settings live in etcd, every material change is audited, and the operator console sits behind Cloudflare Access.

Platform and tooling.

I built Rust services with fixed-point maths, Kafka for events, Postgres for decisions, ClickHouse for history and etcd for live controls, on Kubernetes with Helm, Terraform and GitHub Actions. A paper-trading service and a synthetic market feed allow the whole stack to be regression-tested without risking funds.

Outcome

I co-founded and built the platform from inception, owning architecture, backend and delivery end to end. It is in production.

Read the documentation

Technologies used

RustKafkaPostgreSQLClickHouseetcdParquetHyperliquidKubernetesTerraformGitHub ActionsHelmGrafanaPrometheusNext.jsReactDescopeCloudflareDocker

Data volumes & scale

4 data stores
1 exchange boundary
~12 services
79.7M trade records (dev environment)