Dark rock formations lit by a burning orange horizon

SENTRA®

An AI-based cyber threat detection framework — intrusion analysis, malware classification and explainable risk scoring in one analyst workspace.

Network detection Malware detection

Security Research Lab
Threat Intelligence Unit
Batch inference · CSV

01 Ingestion
02 Feature engineering
03 Model inference
04 Reporting

Project overview

Two detection engines, one analyst surface

SENTRA unifies network intrusion detection and malware family classification behind a single dashboard. Analysts upload batch CSV exports, the detection service scores every record, and the interface turns raw model output into prioritised, explainable findings.

Detection modules

12+

Result surfaces

Reusable components

REST

Backend coupling

Key features

Built for triage, not for demos

Flow-level intrusion detection

Classify network flows into benign traffic or attack families such as DDoS, port scanning and brute force.

Malware family classification

Map static and behavioural features of samples to known malware families with per-record confidence.

Confidence & risk scoring

Every prediction carries a calibrated confidence score and a derived risk level for triage ordering.

Visual analytics

Distribution charts, summary statistics and per-record tables make batch results readable at a glance.

No data retention

Files are streamed to the detection API on demand. The interface keeps nothing in local storage.

API-ready frontend

All result surfaces are typed placeholders wired to a single integration point per detection page.

Technology stack

Frontend today, FastAPI tomorrow

The interface is fully decoupled: swapping placeholder state for live responses touches a single handler per detection page.

React 19 + TypeScript

Component-driven UI layer

FastAPI

Detection service (external)

Pandas / Scikit-learn

Feature pipeline & classical models

TensorFlow / PyTorch

Deep sequence & CNN classifiers

Workflow

How a batch moves through the framework

Upload

Analyst uploads a preprocessed CSV export of flows or samples.

Validate

Format, encoding and schema checks run before anything is sent.

Infer

The detection API scores each record with the trained model.

Report

Predictions, confidence, risk and recommendations are rendered.

Detection modules

Choose an engine to start

Network Intrusion Detection

Upload flow-level CSV exports and classify traffic into benign or attack categories, with attack distribution, risk scoring and record-level output.

Open network detection

Malware Detection

Upload extracted sample features and predict malware families with confidence, severity, recommended containment steps and per-sample results.

Open malware detection