10decodersAML Screening Fast Enough to Feel Invisible
A regulated fintech offering cross-border money transfer, currency exchange, e-wallet, and payment gateway services needed compliance to move as fast as its payments. 10Decoders built a real-time anti-money-laundering screening engine that checks every transaction against sanctions, PEP, watchlist, and adverse media data — and clears it before the payment settles.
Compliance That Keeps Up With the Payment
Every transaction carries a hidden question: is this payment connected to a sanctioned entity, a politically exposed person, a watchlisted individual, or a party named in adverse media? Regulators expect that question answered before funds move — customers expect it answered instantly. Historically, screening systems resolved that tension by making customers wait: batch runs, overnight reconciliation, and manual review queues that could delay a legitimate payment by hours or days.
10Decoders engineered a screening layer that sits directly in the transaction path and checks every payment against sanctions, PEP, watchlist, and adverse media data in a single automated pass — sustaining 200 screening requests per second so compliance rigor no longer costs the customer time.
The Screening Stack
Real-Time Processing
Screening runs in-line with the transaction, not as a batch job after the fact.
In-Memory Reference Data
List lookups stay fast enough to fit inside the payment's live processing window.
Fuzzy Name Matching
Handles name variation and transliteration so genuine matches aren't missed.
Risk-Based Alert Scoring
Ranks alerts by actual risk instead of flooding analysts with noise.
Parallelized Multi-Source Checks
Sanctions, PEP, watchlist, and adverse media run together, not in sequence.
Adverse Media Scanning
Continuously scans news and media sources for reputational risk static lists miss.
Screening had to happen in milliseconds, not minutes.
The Challenge
- Sanctions, PEP, watchlist, and adverse media checks ran as separate, sequential lookups.
- Batch and overnight reconciliation delayed legitimate payments by hours or days.
- Instant payment rails left no real time budget for screening at all.
- Rigid list-matching produced false positives that buried genuine alerts in noise.
Our Approach
- Unified all four sources into a single automated screening pass per transaction.
- Moved screening in-line with the transaction, eliminating batch runs entirely.
- Engineered the pipeline to return a decision inside the payment's live processing window.
- Added fuzzy matching and risk-based scoring to cut false positives and surface real risk.
The Solution
Every transaction is matched, scored, and cleared — or flagged for review — before the payment moves to its next step. The system evaluates each request in parallel against sanctions lists, PEP registries, global watchlists, and adverse media sources in a single automated pass, rather than as four separate lookups.
Day to day, compliance teams see a clear, flag, or block decision returned before settlement. Fuzzy name-matching and transliteration handling keep false positives down, so analysts spend their time on alerts that actually matter — and throughput scales with transaction volume as the business grows.
- 1Screens every transaction against 4 data sources in one automated pass.
- 2Returns a clear, flag, or block decision before settlement.
- 3Fuzzy matching and transliteration handling cut false positives.
- 4Sustains 200 requests per second under live transaction load.
From Transaction to Decision
Sanctions, PEP, Watchlists & Adverse Media
Every transaction and customer record streams in against four continuously updated risk sources.
Parallel Screening & Risk Scoring
Fuzzy name-matching, transliteration handling, and risk-scoring logic run across all sources in one automated pass.
Clear, Flag, or Block
A decision returns inside the transaction's live processing window, before funds settle.
Built for Speed and Scrutiny
Single Unified Pass
Runs sanctions, PEP, watchlist, and adverse media checks together instead of as four separate lookups.
In-Line, Not Batch
Screening sits directly in the transaction path instead of running after the fact.
Fuzzy Name Matching
Handles name variation and transliteration so genuine matches aren't missed.
Risk-Based Alerting
Prioritizes alerts by actual risk instead of flooding analysts with noise.
In-Memory Reference Data
Keeps list lookups fast enough to fit inside the payment's processing window.
Built to Scale
Throughput scales with transaction volume without a proportional rise in headcount.
What Changed
What This Engagement Delivered
Delivery Quality
- Deployed and operating in the client's production environment.
- Sustains 200 screening requests per second under live transaction load.
- 10Decoders' first engagement of this kind — no prior comparable build.
Cost & ROI
- Risk scoring reduces false positives, lowering manual analyst review load.
- Throughput scales with transaction volume without proportional headcount growth.
- Replaces batch reconciliation cycles with continuous real-time processing.
IP & Security
- Not patentable, but the screening module is designed to be reusable across future fintech engagements.
- Built as a core, reusable 10Decoders capability rather than a one-off custom job.
- Produces a defensible, auditable screening trail for regulators and auditors.
Impact Across the Business
Fewer False Positives, More Signal
Risk-scored, fuzzy-matched alerts let analysts focus on genuine risk instead of chasing noise.
Payments That Feel Instant
Screening completes before settlement, so legitimate transactions move without added delay.
Compliance as a Differentiator
Real-time, auditable screening turns a regulatory obligation into a competitive advantage.
