#Fraud
Fraud detection and prevention require real-time analysis of transaction patterns and identity signals. Posts cover anomaly detection for financial fraud, AML compliance engineering, and the multi-layered controls that protect financial platforms from fraudulent activity.
10 posts tagged with fraud. ← All posts
When fraud labels are scarce or the attack is brand new, unsupervised anomaly detection buys you a signal before you ever have a training set.
When you have few or no labels: unsupervised anomaly detection.
Why accuracy lies, how to train on a needle-in-a-haystack, and picking a threshold by the money it saves.
Fraud is a needle-in-a-haystack problem.
How passive device and behavior signals become a trust score for auth and fraud — without turning into a surveillance liability.
How device signals and behavioral biometrics build a trust signal for auth and fraud, with privacy and false-positive trade-offs.
Why per-transaction scoring misses organized fraud, and how modeling accounts, devices, cards, addresses, and IPs as a graph catches the whole ring.
Modeling accounts/devices/payments as a graph, connected-component and community detection, shared-attribute linking, and analyst review.
How to combine a deterministic rules engine with an ML risk model in a single decision: rule precedence, score bands, shadow mode, and champion/challenger evaluation with feedback labels.
Teaches how to combine a deterministic rules engine with an ML risk model in one decision path: rule-precedence and overrides, model score bands, shadow mode, and champion/challenger evaluation with feedback labels.
How a risk-driven layer escalates from silent approval to OTP, biometric, or 3DS challenge — holding a pending-challenge state and resuming the original transaction once the customer clears it.
Teaches how to build risk-based step-up auth: an orchestration layer that escalates from silent to OTP/biometric/3DS challenge based on risk signals, with pending-challenge state and resumable transaction context.
Building low-latency sliding-window counters, entity-keyed aggregates, and an online feature store that stays consistent with its batch-computed twin.
Teaches how to build low-latency velocity checks: sliding-window counters and aggregates in an online feature store, entity keys (card/device/IP), and consistency between real-time and batch-computed features.
When software holds the card and clicks "buy," the old questions — was this the cardholder, did they mean to, who pays if not — all get harder to answer.
The new fraud surface: prompt injection turning a shopping agent into an attacker's buyer, hijacked delegated credentials (Visa saw ~450% more dark-web 'AI Agent' chatter in H1 2026), and disputes when an agent acted on…
Borrower onboarding is the most fraud-prone moment in a P2P platform. The shape that worked: deterministic KYC, parallel bureau pulls with fallback, real-time fraud signals, and a maker-checker approval for every disbursement.
Two signals do most of the work for detecting compromised sessions: impossible travel between consecutive logins, and credential-stuffing density across an IP range. The Go implementation.
All posts on this site are written by Pratik Dhanave, an Agentic AI Architect with 7+ years building production distributed systems, multi-agent AI platforms, and cloud-native infrastructure. About the author → Each article includes working code, architecture diagrams, and references to the specific frameworks and standards discussed. Browse all posts or explore related topics using the tag cloud above.