FintechAI ServicesStartup

AI Fraud Detection for a Neobank Startup

Graph + behavioral fraud scoring that cut false declines while catching account-takeover rings for an early-stage neobank.

AI Fraud Detection for a Neobank Startup

Key Details

ChallengeRules-only fraud checks blocked good users and still missed coordinated account-takeover attempts.
SolutionA real-time fraud scorer combining device, velocity and graph features with an investigator console.
TechnologiesPython, Kafka, Redis, PostgreSQL, AWS

Technologies used

Python Apache Kafka Redis PostgreSQL AWS Docker

Client background

A neobank startup shipping cards and P2P transfers hit a wall: static rules either annoyed legitimate users with declines or let mule rings through. Fraud ops lived in Slack threads with no shared case model.

Key challenges

  • False declines hurt activation and support costs for a brand still earning trust.
  • Coordinated rings shared devices and funding patterns rules could not see.
  • No real-time feature path — batch jobs were too slow for card authorizations.
  • Investigators lacked a prioritized queue with evidence packs.

Product screenshots

Dark purple investigation console with a force-directed link graph of accounts, devices, IP addresses and payees around a flagged account, a risk-score card listing the model's top signals, and a case timeline with freeze and release actions.
Fraud investigation graph: a flagged transfer is linked to the accounts, devices and payees it shares, so the analyst can see the ring the model picked up.
Light lilac banking app shown on three phones: a push notification about a paused transfer, a 'Was this you?' screen with transfer details and yes or no buttons plus face confirmation, and a confirmation screen showing the card frozen and next steps.
Customer-side step-up: when a transfer looks risky the app pauses it and asks 'Was this you?', then confirms or freezes the card in one tap.

What we built

  • Streaming features for device fingerprint, velocity and peer-graph risk.
  • Online fraud score service with kill-switches and reason codes.
  • Investigator console with case notes, linked entities and disposition outcomes.
  • Feedback loop so confirmed fraud retrains weekly challenger models.

Project team: 6 engineers across AI/ML, backend and product engineering — delivery over 14 weeks for a startup team that needed to ship, not slide decks.

Architecture design

Neobank
Customers
Cards & P2P transfers
Card authorisations
Real-time
Devices & sessions
Fingerprints
Investigators
Fraud ops
Applications
Investigator console
Prioritised queue
Case notes
Linked entities
Dispositions
Outcomes
AI & ML
Online fraud score
Reason codes
Kill-switches
Safe rollback
Peer-graph risk
Linked rings
Challenger retraining
Weekly
Data & streaming
Streaming features
Kafka
Velocity & device signals
Real-time
Feature cache
Redis
Case & feedback store
PostgreSQL
Cloud & data services
Streaming
Kafka
Feature streams
Redis
Feature cache
Data
PostgreSQL
Cases & feedback
Cloud
AWS
Hosting
Docker
Containers
Technology stack
Backend & APIs
Python
Data
KafkaRedisPostgreSQL
Cloud & DevOps
AWSDocker

How we delivered

01

Baseline

Measured false decline vs. fraud loss and mapped the hottest attack patterns.

02

Features

Stood up streaming feature jobs with latency budgets for auth time.

03

Score

Shipped v1 scorer with human review for mid-risk band.

04

Operate

Trained fraud ops and locked SLAs for model and rule changes.

Business impact

-35%False declines
+2.1×Fraud catch rate
<80msScore latency

Services used

Industries

We empower businesses with AI, ML, and data solutions.

Response within 1 business day. You will hear from an engineer, not a salesperson.

Contact Us →