Reconcile complex datasets, investigate discrepancies, and produce verifiable evidence with a lightweight, high-performance Rust-native engine.
Transaction exports, settlement files, and banking ledgers often diverge. Asynchronous batch windows, format differences, and micro-deltas multiply. Watch how records fragment across operational systems:
Watch multiple abstract data streams enter the central engine, undergo composite-key alignment, and branch cleanly into verified matches and auditable exception artifacts.
Rather than emitting monolithic CSV diff dumps that crash spreadsheets, Rift isolates every mismatch into a typed, structured evidence packet with cryptographic hash provenance.
Value delta -80.00 detected in post-processing batch. Attributed to promotional MDR rebate.
{
"status": "modified",
"record_id": "C24580",
"field": "amount",
"before": "1230.00",
"after": "1150.00",
"delta": "-80.00",
"rule_candidate": "RULE_MDR_80_REBATE",
"confidence": "exact",
"hash": "sha256:7f83b1657ff1fc53b92dc18148a1d65dfc2d4b1fa3d677284addd200126d9069"
}Rift is engineered as a compiled native binary. No runtime JVM warmup, no Python GIL lockups, and no permanent application tier required.
Rift operates as a focused read-only analysis layer designed to complement existing settlement systems and data lakes without requiring architectural overhauls.
POS gateways, settlement batches, core banking transactions, and staging tables. Ingested via native Parquet, JSONL, CSV, TSV, or PostgreSQL table/view readers.
Runs as an ephemeral container, scheduled job, or CLI binary. Executes dual-engine reconciliation (in-memory or external sort/merge) within user-configured sorter budgets.
Generates cryptographic SHA-256 execution hashes, exact discrepancy classifications (Modified, Missing, Extra), and verified PGDS candidate explanations.
Settlement desks, dispute resolution queues, and SIEM monitoring pipelines ingest structured results directly to resolve exceptions in seconds.
Runs as an ephemeral container, scheduled job, or CLI binary. Executes dual-engine reconciliation (in-memory or external sort/merge) within user-configured sorter budgets.
Reconcile millions of daily card, UPI, and POS authorizations against switch clearing batches.
In high-velocity merchant acquiring, asynchronous settlement batches diverge constantly. Rift isolates fee micro-deltas, timing lags, and missing refund credits in seconds.
rift compare pos_tx.jsonl switch_batch.parquet -s payment_schema.json -k rrn,auth_code
Measured under isolated operating system kernel accounting (/usr/bin/time -l) on Apple Silicon M4. Sorter allocation budget strictly separated from process resident memory (RSS).
Empirical benchmark harness metrics measured via isolated OS kernel accounting (/usr/bin/time -l).
| Workload | Engine Mode | Runtime | Peak RSS | Throughput | Notes & Conditions |
|---|---|---|---|---|---|
| 100,000 Records (Parquet) | In-memory (Optimized) | 0.113s | 42.1 MB | 884,955 rows/s | 2.83× speedup over un-warmed baseline (0.320s median). Warm protocol. |
| 1,000,000 Records (Parquet) | In-memory Reference | 4.05s | 2,803 MB | 246,913 rows/s | Baseline in-memory key-indexed hash table; full dataset retained in RAM. |
| 1,000,000 Records (Parquet) | External Sort/Merge (64MB) | 2.03s | 98.0 MB | 492,610 rows/s | 64MB sorter budget; streaming chunk spill to NVMe; 96.5% lower RSS. |
| 5,000,000 Records (Parquet) | External Sort/Merge (64MB) | 10.19s | 97.3 MB | 490,677 rows/s | Constant bounded memory footprint despite 5× dataset volume scaling. |
As global payment processors, digital banks, and autonomous AI data platforms scale, the economic damage of undetected data divergence escalates exponentially. Explore Eagle Eye’s long-term commercialization strategy, research milestones, and institutional investment thesis.
Benchmark RIFT against your existing reconciliation methods under a read-only, non-disclosure evaluation agreement.