live · base mainnet · usdc

Real-time financial intelligence,
extracted from raw blockchain data

OmniSight ingests every USDC transfer on Base Mainnet, decodes the raw EVM payload, and scores wallet behaviour with a trained anomaly model — all visible below, updating right now.

API docs GitHub
omnisight — whale-alerts.live
connecting
Transfers tracked
≥$50K · rolling sample
Volume tracked
USD
Largest transfer
USD
Ingestion latency
chain → dashboard
Live whale transfers
BlockTxSenderReceiverAmountAge
Loading live feed…

Eight stages, ledger to risk score

Every layer designed, deployed, and maintained independently. No managed services — just the raw stack.

01
Base Mainnet — JSON-RPC via Alchemy
Raw event logs pulled directly from the chain. No intermediary indexer.
web3.py
02
Airflow DAG — 120s cadence
Incremental checkpointing, idempotent inserts, self-healing on block failures.
Airflow
03
EVM decoder
32-byte topic fields decoded into wallet addresses, 6-decimal USDC normalisation.
Python
04
PostgreSQL — partitioned by block range
Sub-second range queries at scale. Unique constraint on transaction hash.
Postgres 15
05
Feature engineering
Transaction count, total volume, average size — per wallet, computed on demand.
SQL
06
RobustScaler + Isolation Forest
Unsupervised anomaly detection. Median/IQR scaling resists whale-wallet outliers.
Scikit-Learn
07
FastAPI gateway
Authenticated and public rate-limited endpoints. Async connection pooling.
FastAPI
08
Nginx + TLS — always on
systemd-managed, survives reboots, Let's Encrypt auto-renewal.
OCI

Why this matters

On-chain volume keeps growing faster than the tooling to monitor it. Most teams still rely on manual lookups or closed enterprise platforms that take weeks to integrate.

OmniSight is the alternative: a single engineer, an open codebase, and infrastructure that costs nothing to run — proof that real-time risk intelligence doesn't require a six-figure contract or a team.

Operational facts

  • 393,000+ clean records, zero duplicates enforced by DB constraint
  • 9,410 wallets with 2+ transactions, eligible for ML scoring
  • Model , retrained on wallet profiles
  • API survives reboots — systemd managed

Three endpoints, fully documented

Full interactive Swagger UI at omnisight.ericdiamason.tech/docs

GET/api/v1/metrics/whale-alerts
GET/api/v1/public/wallet-risk?wallet_address=0x…
GET/api/v1/predict/wallet-risk — requires X-API-Key

Wallet risk explorer

Run any Base Mainnet wallet through the live Isolation Forest model. Rate limited to 10 requests/minute, no key required.

Enter a wallet address above, or try the example — a wallet with 9,119 transactions and $888M in volume.