Skip to content

Latest commit

 

History

6 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 

Repository files navigation

Onyedika Christopher Agada

Cybersecurity and machine-learning researcher building evidence-grounded systems for trustworthy security decisions.

SecOpsAI · Documentation · Research blog · ORCID · Publication · LinkedIn

Profile

I am a cybersecurity researcher and engineer based in Istanbul, Türkiye. I hold an MSc in Cybersecurity from Üsküdar University, completed with High Honour, and a BSc in Biochemistry from the University of Port Harcourt.

My work combines malware analysis, software-supply-chain security, machine learning, security operations, and the design of systems that keep evidence, uncertainty, and human responsibility visible. I am the sole creator of SecOpsAI, a local-first platform for security telemetry, investigations, package research, detection learning, and controlled publication.

Research interests

  • Evidence-grounded generative and agentic AI for cybersecurity
  • Malware classification and software-supply-chain threat analysis
  • Causal, explainable, and uncertainty-aware machine learning
  • Knowledge representation, provenance, and formal reasoning for reliable systems
  • Human oversight, abstention, auditability, and rollback in automated decisions
  • Privacy-conscious and local-first security architecture

These are active research directions. I distinguish them from capabilities already validated through completed experiments or publications.

SecOpsAI

Component Purpose
SecOpsAI Core Evidence-first triage, package research, asset graphs, detection workflows, agent telemetry, and guarded automation.
Mission Control Operator interface for findings, investigations, research cases, automation, disclosure, and reviewed publication.
SecOpsAI Edge Local sensor for authorised asset discovery, risky-service detection, change tracking, and normalised Core ingestion.
Research workflow Safe artefact intake, static analysis, package comparison, evidence provenance, IOC review, disclosure controls, and publication gates.

SecOpsAI treats model output as analytical assistance, not evidence. High-impact actions remain bounded by explicit approval, audit records, and reproducible inputs.

Selected work

Project Focus
Malware Detection on an Optimised Edge Device Transfer learning, model optimisation, ONNX inference, and low-latency Android deployment.
Deep Learning Malware Classification CNN-based malware-family classification and experimental evaluation.
AI-Powered SOC Agent Bounded natural-language security analysis and structured incident reporting.
AWS Bedrock RAG Project Terraform-deployed retrieval workflow using Bedrock, Aurora PostgreSQL with pgvector, S3, and scoped IAM.
SecOps Autoresearch Controlled detection optimisation with evaluation, feedback, and rollback-oriented workflows.

Publication and thesis

Agada, O. C. (2025). “From Training to Edge Inference: A Transfer Learning Pipeline for Low-Latency Malware Classification on Snapdragon-Enabled Samsung Galaxy S23 Ultra (Android 13).” Journal of Scientific Reports, 11(1), 51–64. DOI: 10.58970/JSR.1135

My MSc thesis, “A Comparative Analysis of Transfer Learning Approaches for Malware Classification Using Convolutional Neural Networks,” compared transfer-learning architectures on malware-family datasets using accuracy, precision, recall, F1-score, AUC, confusion matrices, class-imbalance analysis, and generalisation checks. YÖK National Thesis Center Reference No. 10700043.

Working principles

  • Preserve the source and provenance of every material claim.
  • Separate observed facts, reasoned inferences, model suggestions, and unsupported conclusions.
  • Never execute untrusted packages on an operator or application host.
  • Require explicit approval for sandbox execution, disclosure, mitigation, and publication.
  • Evaluate detection changes on holdout data before controlled activation.
  • Record negative results, limitations, tool versions, and recovery paths.

Tools

Python · FastAPI · PyTorch · TensorFlow · ONNX Runtime · JavaScript · Bash · SQL · PostgreSQL · Docker · Terraform · AWS · Google Cloud · Cloudflare · GitHub Actions

Contact

For research or security reports, include reproducible evidence, affected versions or systems, collection dates, and the limits of the available evidence.

About

Cybersecurity and machine-learning research, SecOpsAI, malware analysis, and trustworthy agentic AI.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors