train · ok
threat · NOMINAL
07:51:55
online · Colombo, LK · 07:51:55 UTC+05:30 · uptime 2y 6m

Thathsara
Madusha.

Building full-stack systems and the DevOps plumbing that keeps them honest — from Kubernetes clusters to AI-powered platforms.

studying · BSc CS (Hons.) @ UCSC researching · DDoS-aware autoscaling building · homelab + open source GPA · 3.66 · Director’s List ×2
[07:00] homelab k8s cluster health: all nodes ready, uptime 47d [07:07] research ddos-autoscaling: methodology draft in review [08:14] github smile: PR merged — AI recommendation engine v2 [08:21] ucsc GPA: 3.66 — Director’s List 2023 & 2024 [09:28] research syn-dpi-bypass: feasibility analysis 60% complete [09:35] github pixel-ai: Ballerina API exam module deployed [10:42] homelab grafana: 0 critical alerts, prometheus scrape 30s [10:49] coffee v60 #2 brewed, focus level nominal [11:56] homelab k8s cluster health: all nodes ready, uptime 47d [11:03] research ddos-autoscaling: methodology draft in review [12:10] github smile: PR merged — AI recommendation engine v2 [12:17] ucsc GPA: 3.66 — Director’s List 2023 & 2024 [13:24] research syn-dpi-bypass: feasibility analysis 60% complete [13:31] github pixel-ai: Ballerina API exam module deployed [14:38] homelab grafana: 0 critical alerts, prometheus scrape 30s [14:45] coffee v60 #2 brewed, focus level nominal

// reach

email  thathsaramadhhusha@gmail.com
github @nova9
twttr  @thathsara_m
pgp    A5F8 2C0D 9E11 4B73
// PSST
there are a few hidden things on this page.
try typing help, secrets, or the konami code. ↑↑↓↓←→←→ba

about

whoami.md active

Software engineer at UCSC building full-stack systems and the DevOps plumbing that keeps them honest — from Kubernetes clusters to AI-powered platforms.

I’m a software engineer at the University of Colombo School of Computing. I work across the stack: backend services in Go, React frontends, container orchestration with Kubernetes, and cloud infra on AWS and GCP.

I’ve shipped production features as a full-stack intern at WSO2, built freelance Laravel apps from 2022–2024, and run a self-hosted homelab exposed to the public internet. I like building things that actually run and don’t fall over.

Currently researching DDoS-aware autoscaling in cloud environments and data exfiltration via TCP SYN packets. Previously: Organizing Committee VP at AIESEC Insight 2024. I hold the Director’s List Award for academic excellence in 2023 and 2024.

✦ Otherwise: homelab tinkering at odd hours, a half-finished research paper about TCP SYN packets, and coffee. Open to internships, weird collaborations, and projects that ship.

profile.jsonverified
nameThathsara Madusha handle@nova9 roleSoftware Engineer & DevOps Practitioner locColombo, LK · UTC+05:30 statusshipping stack gotsphppy domains webdevopsai pgpA5F8 2C0D 9E11 4B73
        ▒▒▒▒▒▒▒▒▒▒        
     ▒▒░░░░░░░░░░░░▒▒     
   ▒▒░░░░░░░░░░░░░░░░▒▒   
  ▒░░░░▓▓░░░░░░░░▓▓░░░░▒  
  ▒░░░░░░░░░░░░░░░░░░░░▒  
  ▒░░░░░░░░░░██░░░░░░░░▒  
  ▒░░░░░░██████████░░░░▒  
   ▒▒░░░░░░░░░░░░░░░░▒▒   
     ▒▒░░░░░░░░░░░░▒▒     
    ▓▓▓▓▓░░░░░░░░▓▓▓▓▓    

projects.ml

projects/ 5 active · 12 archived
01

homelab

secoss

Self-hosted infrastructure lab running production-grade services on commodity hardware, exposed to the public internet. Kubernetes for container orchestration, Nginx as reverse proxy with TLS termination. Full observability via Grafana and Prometheus dashboards.

lang · YAML, Shell shipped · 2024 stack · Kubernetes · Nginx · Grafana +3
// live
cluster: Kubernetes
proxy: Nginx + TLS termination
observability: Grafana + Prometheus
exposed: public internet
↗ github
02

smile

mloss

Volunteer management platform with an AI-powered event recommendation engine. Matches volunteers to opportunities using OpenAI embeddings + cosine similarity over Google Maps data. Real-time messaging between volunteers and organisers, plus AWS Rekognition for identity verification.

lang · PHP, JavaScript shipped · 2025 stack · Laravel · Livewire · Volt +3
// live
rec engine: cosine similarity
embeddings: OpenAI
verification: AWS Rekognition
comms: real-time messaging
03

pixel-ai

mloss

Comprehensive e-learning platform with AI-driven content generation. React/TypeScript frontend backed by a Ballerina microservices API covering user management, content delivery, exam management, and progress tracking. OpenAI for generation, MySQL for persistence, AWS SES for auth.

lang · TypeScript, Ballerina shipped · 2025 stack · React · TypeScript · Ballerina +3
// live
services: user, content, exam, progress
auth: AWS SES
content gen: OpenAI
API: RESTful modular
04

talent-pipeline

mlsec

Cloud-native recruitment platform that processed 12,000+ résumé PDFs using OpenAI. Deployed on Kubernetes via Terraform with a React/Laravel stack. AI-driven features connecting candidates, employers, and recruiters for career guidance and streamlined hiring at scale.

lang · TypeScript, PHP shipped · 2025 stack · React · Laravel · Kubernetes +3
// live
resumes processed: 12,000+
infra: Kubernetes via Terraform
stack: React + Laravel
features: AI matching, career guidance
↗ github
05

launchpad

osssec

Internship management system built on a custom MVC PHP framework written from scratch — routing, middleware, and ORM included. Established automated test coverage with Pest for unit tests and Selenium for end-to-end browser testing.

lang · PHP shipped · 2024 stack · PHP · Custom MVC · Pest +2
// live
framework: built from scratch
routing + ORM: custom
unit tests: Pest
E2E tests: Selenium

open source

github.com/nova9 live data
987
contributions this year
25
public repositories
13
followers
6
top repos shown

writing

blog/access.log rss available
# tail -f ~/blog/access.log   ·   8 entries
2026-05-01 STDY
DDoS-Aware Autoscaling: distinguishing legitimate vs. malicious traffic
Research notes on EDoS attacks and resource allocation under adversarial load.
cloudsecurityresearch
2025-12-10 HACK
Bypassing Deep Packet Inspection via Data-in-SYN Packets
Technical feasibility and economic viability of embedding data in TCP SYN packets.
networkingsecurityresearch
2025-09-14 NOTE
Building a volunteer matching engine with OpenAI embeddings
Cosine similarity over embeddings + Google Maps for event-volunteer matching.
mlembeddingslaravel
2025-08-20 INFO
Processing 12,000 résumés with OpenAI on Kubernetes
Lessons from building a scalable AI recruitment pipeline on a student budget.
k8sopenaicloud
2025-07-05 HACK
Ballerina as a backend language: a React dev’s perspective
What it’s like to wire up a Ballerina microservices API for the first time.
ballerinabackendapi
2025-02-14 NOTE
WSO2 internship: shipping persistent storage across 6 codebases
Full-stack feature delivery through senior review, CI, and integration validation.
goreactwso2
2024-11-03 INFO
Building a PHP MVC framework from scratch
Routing, middleware, and ORM — what I learned skipping the framework.
phpmvcarchitecture
2024-06-20 HACK
Homelab on commodity hardware: Kubernetes, Nginx, and public internet exposure
Running production-grade infra at home without spending production-grade money.
k8shomelabdevops
# [end of log] — older entries: archive/

stack

skills.toml self-reported, biased

Languages

Go
2024
JavaScript / TypeScript
2022
PHP
2022
Python
2023
C / C++
2023
SQL
2022

Web

Laravel
2022
React
2023
Vue.JS
2023
NodeJS
2022
Livewire / Volt
2024
Ballerina
2025

Tools & DevOps

Docker
2023
Kubernetes
2024
Terraform
2025
GitHub Actions
2023
Grafana / Prometheus
2024
Git
2022

Platforms

Amazon Web Services
2024
Google Cloud Platform
2025
Nginx
2024
MySQL / PostgreSQL
2022
Linux / Shell
2022
Selenium / Pest
2024

operating philosophy

ship it, then observe it. I'd rather have a Grafana dashboard that tells me something broke than an architecture diagram that says it won't. I'd rather write the migration than the slide deck. The most satisfying work is code that runs in prod without anyone noticing.

# things I will fight you on:
frameworks exist to be replaced · a custom ORM teaches you more than using one · observability is not optional · Kubernetes is worth the pain.

timeline

git log --author=thathsara 3 entries · 1 current
2025 — 2026

Intern Software Engineer @ WSO2 LLC

Designed and shipped a full-stack feature for persistent storage attachment to cloud applications. Backend services in Go handling volume lifecycle management and deduplication. React/TypeScript marketplace UI for shared storage resources. Coordinated across 6 codebases and 3 teams.

PRs: 7 mergedlang: Go + React/TSscope: 6 codebases
2023 — now

BSc Computer Science (Hons.) @ UCSC, Colombo

GPA 3.66. Director’s List 2023 and 2024. Researching DDoS-aware autoscaling and TCP SYN data exfiltration. AIESEC Organizing Committee VP for Insight 2024.

GPA: 3.66awards: Director's List ×2research: 2 ongoing
2022 — 2024

Freelance Web Developer @ Remote

Delivered Laravel-based CMS and web applications for multiple clients. Owned backend architecture, deployment, and ongoing maintenance. Improved client search visibility through SEO best practices and structured data markup.

stack: Laravel + PHPfocus: CMS & web appsextras: SEO + performance
2020

$ git init

First commit. A PHP script to scrape exam results. Lost the repo, kept the habit.

now

~/.now updated may 27 2026

currently running PID 0–6

pidcmdcpumem
0x01 research --topic ddos-autoscaling --status ongoing 62% 2G
0x02 research --topic syn-packet-dpi-bypass --status ongoing 45% 1G
0x03 study UCSC Year 3 Semester 1 80% 4G
0x04 homelab --cluster k8s --status running --uptime 47d 12% 512M
0x05 read Computer Networking: A Top-Down Approach 3% 64M
0x06 brew --method v60 --beans local-roast 70%

Pretraining a 350M-param policy model on a year of CloudTrail traces. Writing chapter 3 of Verifiable ML in Regulated Clouds (early 2027, manning). On-call primary this rotation; pager has been quiet — I’d like to keep it that way.

reading curr / done / next

Computer Networking: A Top-Down Approach
Kurose & Ross
cur
Cloud Native DevOps with Kubernetes
Arundel & Domingus
cur
The Phoenix Project
Gene Kim et al.
done
Designing Data-Intensive Applications
Martin Kleppmann
todo
Clean Architecture
Robert C. Martin
todo

inbox 4 unread

Semester results: GPA 3.66 maintained
from ucsc-portal
LO 2d
homelab CI: cluster health check green
from github-actions
LO 4h
DDoS autoscaling: feedback on methodology draft
from research-supervisor
MD 1d
Daily digest: cs.NI + cs.CR (18 papers)
from arxiv
LO 6h
thathsara@portfolio:~$