Kampala, Uganda · Open to security work

I break systems to them.

Cybersecurity expert and software engineering student testing networks, writing exploit proof-of-concepts, and building systems that hold.

See my work Check my CV
Portrait of Atigo Paul Odoch in front of a brick church courtyard
ATIGO PAUL ODOCH SEC / 2026

Built for the offence,
trusted on the defence.

I work from packet capture to production code. My security research is public, reproducible and written to help defenders close the gaps.

4 Public repositories
11 Anthropic certificates
06 CVE proof-of-concepts
04 Core disciplines

I am a cybersecurity analyst, network engineer and software engineering student at Mbarara University of Science and Technology. I run vulnerability assessments, perform ethical hacking and build secure systems end to end.

My public research focuses on SMB and Windows attack paths: Samba pre-auth RCE, EternalBlue, SMBGhost and PwnKit. I also work the other side: Django and Node services, hardened APIs and monitored networks.

Currently an AI engineering intern at Deep Shift AI, training and evaluating models for Ugandan contexts, and leading the cybersecurity division at MUCOSA. I also take on general software design and build web platforms through GrooveStreet Software.

Languages: English, Langi, Luganda, Lusoga, Kiswahili, Runyankore. Off the clock: music, art, video games, workouts, rugby and vibe coding.

What I do

Four disciplines, one goal: systems that survive contact with an attacker.

  1. 01

    Offensive security

    Penetration tests, exploit proof-of-concepts and privilege escalation across SMB, Windows and Linux targets. Findings arrive with reproduction steps, not just a score.

    NmapBurp SuiteMetasploitPoC development
  2. 02

    Network engineering

    Segmented architectures, firewall policy and traffic analysis. I design networks that are observable by default and painful to move through laterally.

    TCP/IPRoutingFirewallsWireshark
  3. 03

    Secure software

    Web applications and APIs built against the OWASP Top 10, with auth, input handling and logging treated as product features rather than a final audit.

    PythonDjangoNode.jsAPI security
  4. 04

    AI & automation

    LLM agents, tool use and secure automation. Certified across Claude, Google Vertex AI and Amazon Bedrock, applied to security workflows and product delivery.

    ClaudeVertex AIAmazon BedrockAgents

Selected projects

Public research and hands-on builds from my security GitHub.

Research · Featured ↗

SMB Exploit Lab

A working collection of SMB and local privilege escalation proof-of-concepts, verified against Metasploitable 2 and 3. Standard library Python, clear scope, reproducible labs.

  • CVE-2007-2447 Samba pre-auth RCE
  • CVE-2009-3103 Windows SMBv2 client crash
  • CVE-2010-2075 UnrealIRCd backdoor
  • CVE-2017-0143 EternalBlue / MS17-010
  • CVE-2020-0796 SMBGhost RCE
  • CVE-2021-4034 PwnKit local root
PythonSMBExploit devLinux
GitHub Public
Windows · PoC ↗

SMBGhost RCE

CVE-2020-0796 proof-of-concept for Windows 10 1903 and 1909. Kernel offset collection, SMBv3 compression and a SYSTEM shell chain.

WindowsKernelPython
GitHub
Web security · Lab ↗

VulnMart

A deliberately vulnerable PHP storefront used to practise web exploitation end to end: auth bypass, injection, broken access control and reporting.

PHPWeb securityOWASP
Fork

Certified across the AI stack.

Eleven completed Anthropic Academy programs covering Claude, agents, cloud deployment and AI fluency. Every certificate is verifiable and linked.

Claude 101 certificate Anthropic Academy

Claude 101

View certificate Claude Code in Action certificate Anthropic Academy

Claude Code in Action

View certificate Introduction to Claude Cowork certificate Anthropic Academy

Introduction to Claude Cowork

View certificate Introduction to subagents certificate Anthropic Academy

Introduction to subagents

View certificate Introduction to agent skills certificate Anthropic Academy

Introduction to agent skills

View certificate Claude with Google Vertex AI certificate Anthropic Academy

Claude with Google Vertex AI

View certificate Claude with Amazon Bedrock certificate Anthropic Academy

Claude with Amazon Bedrock

View certificate AI Fluency: Framework and Foundations certificate Anthropic Academy

AI Fluency: Framework & Foundations

View certificate AI Fluency: AI Capabilities and Limitations certificate Anthropic Academy

AI Capabilities & Limitations

View certificate AI Fluency for students certificate Anthropic Academy

AI Fluency for students

View certificate AI Fluency for Small Businesses certificate PayPal × Anthropic

AI Fluency for Small Businesses

View certificate
CompTIA

Network+ (N10-008)

Certified

Where I've worked

Engineering with security in the requirements, not bolted on after launch.

2026 — Present

AI Engineering Intern

Deep Shift AI · Kampala

  • Train and evaluate AI models across the full pipeline: data curation, prompt design, evaluation and deployment.
  • Ground model behaviour in Ugandan context, language and local use cases.
Current

Team Lead · Cybersecurity Division

MUCOSA · Mbarara University of Science and Technology

  • Lead the student cybersecurity division: hands-on labs, CTFs and security awareness.
  • Coordinate division research, tooling and the training calendar.
2025

Network Engineering Intern

ICT Labs · Ntinda

  • Built practical routing, switching, device configuration and troubleshooting skills from beginner to intermediate level.

Let's secure your next system.

Available for penetration tests, security reviews, network design, secure web builds and general software design. Based in Kampala, working remotely.

Opens your email app with the message ready to send.