Malware Analysis Tool
Static analysis that speaks detection language: PE inspection, YARA matches and MITRE ATT&CK mapping in under thirty seconds.
I research attack surfaces across iOS runtimes, cloud infrastructure, APIs, binaries, and AI agents. The work moves from exploit path to detection, hardening, and a reproducible technical case study.

Offensive security
Amdjed Salhi
Offensive Security Engineer
Software engineering is the base layer; offensive security is the discipline; AI systems are the multiplier.
I work across the boundary between software construction and adversarial analysis: understand the system, identify the trust boundary, prove the path, then ship the control that closes it.
My current focus is mobile and iOS internals, cloud-native attack paths, and AI-assisted vulnerability research. I build the tooling around the research: Frida instrumentation, binary triage, fuzzing harnesses, attack-graph enumeration, runtime telemetry, and reproducible reports.
The operating principle is simple: every offensive finding should produce evidence, a clear impact model, and a remediation path that an engineering team can implement. That is how a lab becomes useful security engineering.
Amdjed Salhi
Offensive Security Engineer
Map identities, privileges, inputs, and execution paths before testing controls.
Prioritise reachable attack chains over isolated scanner output.
Turn repeatable analysis into harnesses, telemetry, tests, and reports.
A finding is incomplete until the defensive change is specific and verifiable.
Every project below answers the same question: can this person make a system that still holds when someone attacks it?
Static analysis that speaks detection language: PE inspection, YARA matches and MITRE ATT&CK mapping in under thirty seconds.
A segmented, instrumented SOC you can attack on purpose: pfSense, Security Onion, Suricata and Zeek with five documented scenarios.
Volatility 3 plus twelve heuristics for process hollowing, injection, credential dumping and rootkits — output as a report a lawyer could read.
A TF-IDF + scikit-learn pipeline that reads urgency, URL anomalies and sender spoofing at 12 ms per message, served as an API any mail gateway can call.
A production commerce, inventory, and mobile ecosystem used as a live testbed for iOS hardening, API exploitation, IAM, and DevSecOps controls.
A full-stack product shipped end to end as my final-year project — and recognised with the 2025 Innovating Project award.
A modular Frida runner for Objective-C and Swift runtime inspection, security-control analysis, and automated data-flow tracing.
An intentionally misconfigured AWS and container range for documenting privilege escalation chains and validating Terraform remediation.
A multi-agent pipeline for decompilation review, high-risk source analysis, CWE mapping, and exploitability scoring.
A crash-triage framework for exposed iOS/macOS XPC services and Mach IPC endpoints, with structure-aware payload generation.
A pipeline that compares pre- and post-patch binaries, isolates silent security fixes, and ranks reachable functions for review.
A vulnerable-versus-hardened Kubernetes range for container escape, sidecar injection, metadata theft, and kernel-level enforcement.
An adversarial testing framework for AI agents, MCP servers, and RAG workflows covering injection, tool poisoning, SSRF, and data exfiltration.
Full case studies — architecture, evidence and the talk track — live on each project page.
All repositoriesFrontier work that extends the shipped portfolio: mobile internals, cloud-native attack paths, and AI systems designed for vulnerability research.
A practical research line covering Mach-O analysis, Objective-C and Swift runtime instrumentation, Keychain and TLS auditing, API exploitation, and XPC/Mach IPC fuzzing.
Intentionally vulnerable AWS and Kubernetes ranges paired with Terraform, OPA, CloudTrail, and eBPF controls. Each exploit chain ends in a tested defensive baseline.
Agentic pipelines for binary patch diffing, decompilation triage, exploitability scoring, prompt-injection testing, MCP tool poisoning, and RAG data-exfiltration research.
MethodEvery research direction is scoped as an engineering system: attack surface, instrumentation, proof, remediation.
One degree, one innovation award, and a stack of practical labs. No certificate theatre — the work has to carry the weight.
Academic record
B.Sc. — conferred 2024
Foundation in software engineering, operating systems, networks, distributed systems, and end-to-end product delivery.
Innovation award
2025
visuflex.online
Recognition for an end-to-end product engineered, deployed, and defended under formal technical review.
Practical record
Hands-on labs, training, and technical work. Issuer documentation and project evidence available on request.
A technical progression from distributed systems and networks into low-level analysis, detection engineering, mobile security, cloud-native attack paths, and AI vulnerability research.
Architecting and cloud-deploying a commerce, inventory, and mobile ecosystem as a live adversarial testbed. Building iOS hardening research, cloud-native attack paths, AI-assisted binary triage, and agent security testing.
Graduated in System Information & Software Engineering. Engineered, deployed, and defended visuflex.online, a full-stack real-time video communication and automated summarisation platform recognised with the Innovating Project label.
Moved from application development into adversarial engineering: network protocols, memory corruption mechanics, Linux and Windows internals, and detection engineering with segmented environments and custom Suricata, Zeek, and YARA rules.
Built full-stack web architectures, smart-contract prototypes, and automated backend services. Established a systems-first workflow for finding architectural bottlenecks and shipping working code.
Darwin XPC fuzzing, cloud-native escape research, and agentic vulnerability pipelines — each paired with reproducible evidence and defensive controls.
Active researchTools are evidence of a method, not a personality. These are the ones I reach for when a system has to work.
Arabic — native · French — professional · English — professional
A live view of the system: what I am building, what I am learning, what is already running, and what I am open to.
Testing iOS storage, TLS, anti-tamper, API authorisation, and cloud delivery controls with code-level remediation.
Modular Frida automation for Objective-C / Swift runtime inspection, cryptographic call tracing, and insecure data-flow auditing.
Developing attack-graph enumeration, Terraform guardrails, binary patch diffing, and structured exploitability scoring.
Mobile security, cloud-native red teaming, AI security, DFIR, and detection engineering.
Open to work. Offensive security, mobile security, cloud-native security, AI security, and vulnerability research roles.
Start the conversationI am open to offensive security engineering, mobile security, cloud-native security, AI security, and vulnerability research roles.
Base
Constantine, Algeria
Local time
--:-- — UTC+1
Status
Available for technical security work — 2026
Currently
Building mobile and iOS security tooling, cloud-native attack-path labs, and AI-assisted vulnerability research pipelines.