-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathindex.html
More file actions
320 lines (285 loc) · 40.9 KB
/
index.html
File metadata and controls
320 lines (285 loc) · 40.9 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<meta name="description" content="Systems builder, AI safety researcher, and adversarial thinker with nearly 45 years across the stack. The failure-first methodology didn't come from papers ...">
<meta name="keywords" content="Adrian Wedd, AI Engineer, Software Architect, Machine Learning, Autonomous Systems, Python, JavaScript, Tasmania">
<meta name="author" content="Adrian Wedd">
<!-- Open Graph / Social Media -->
<meta property="og:type" content="website">
<meta property="og:site_name" content="Adrian Wedd — CV">
<meta property="og:locale" content="en_AU">
<meta property="og:title" content="Adrian Wedd - AI Safety Researcher & Developer">
<meta property="og:description" content="Systems builder, AI safety researcher, and adversarial thinker with nearly 45 years across the stack. The failure-first methodology didn't come from papers ...">
<meta property="og:url" content="https://cv.adrianwedd.com">
<meta property="og:image" content="https://cv.adrianwedd.com/assets/og-image.png">
<meta property="og:image:width" content="1200">
<meta property="og:image:height" content="630">
<!-- Twitter Card -->
<meta name="twitter:card" content="summary_large_image">
<meta name="twitter:title" content="Adrian Wedd - AI Safety Researcher & Developer">
<meta name="twitter:description" content="Systems builder, AI safety researcher, and adversarial thinker with nearly 45 years across the stack. The failure-first methodology didn't come from papers — it...">
<meta name="twitter:image" content="https://cv.adrianwedd.com/assets/og-image.png">
<title>Adrian Wedd - AI Safety Researcher & Developer</title>
<link rel="canonical" href="https://cv.adrianwedd.com">
<!-- Fonts -->
<link rel="preconnect" href="https://fonts.googleapis.com">
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
<link href="https://fonts.googleapis.com/css2?family=Inter:wght@400;600&family=JetBrains+Mono:wght@400&display=swap" rel="stylesheet">
<!-- Favicon -->
<link rel="icon" type="image/svg+xml" href="favicon.svg">
<link rel="icon" type="image/x-icon" href="favicon.ico" sizes="32x32">
<link rel="apple-touch-icon" href="apple-touch-icon.png">
<link rel="manifest" href="manifest.json">
<!-- Security Headers — 'unsafe-inline' in script-src required for pipeline-injected cv-inline-data; all other inline scripts externalised -->
<meta http-equiv="Content-Security-Policy" content="default-src 'self'; script-src 'self' 'unsafe-inline' https://www.googletagmanager.com https://snap.licdn.com; style-src 'self' 'unsafe-inline' https://fonts.googleapis.com; font-src https://fonts.gstatic.com; img-src 'self' data: https:; connect-src 'self' https://api.github.com https://www.google-analytics.com https://analytics.google.com https://px.ads.linkedin.com">
<meta name="referrer" content="strict-origin-when-cross-origin">
<!-- LinkedIn Insight Tag -->
<script defer src="assets/linkedin-insight.js"></script>
<noscript><img height="1" width="1" style="display:none;" alt="" src="https://px.ads.linkedin.com/collect/?pid=8890876&fmt=gif" /></noscript>
<!-- Analytics -->
<script async src="https://www.googletagmanager.com/gtag/js?id=G-ET0FJJS7C7"></script>
<script defer src="assets/analytics-config.js"></script>
<!-- Styles -->
<link rel="stylesheet" href="assets/styles.css">
<!-- Structured Data -->
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Person",
"name": "Adrian Wedd",
"jobTitle": "AI Safety Researcher & Developer",
"description": "Systems builder, AI safety researcher, and adversarial thinker with nearly 45 years across the stack. The failure-first methodology didn't come from papers — it came from coordinating direct actions for Greenpeace's Actions unit against well-resourced opponents where getting the risk assessment wrong had real consequences. That operational security instinct runs through everything: seven years leading cybersecurity, penetration testing, and IDAM for Tasmania's public housing sector; three years of empirical AI red-teaming across 120+ models and 18,000+ adversarial prompts; and production AI systems with hallucination detection and content safety gates built in from the start. Currently freelancing across client delivery (healthcare ecosystems, multilingual sites, ops tooling) and independent AI safety research. AuDHD — the hyperfocus and pattern recognition are features, not bugs. Most interested in where complex systems break and what that reveals about how they were built.",
"url": "https://cv.adrianwedd.com",
"email": "adrian@adrianwedd.com",
"telephone": "+61407081084",
"sameAs": [
"https://github.com/adrianwedd",
"https://linkedin.com/in/adrianwedd"
],
"knowsAbout": [
"Python",
"JavaScript",
"TypeScript",
"PowerShell",
"Large Language Models",
"AI Safety & Evaluation",
"Process Automation",
"Systems Integration",
"Cybersecurity",
"Linux/Windows Server"
],
"address": {
"@type": "PostalAddress",
"postOfficeBoxNumber": "331",
"addressLocality": "Cygnet",
"addressRegion": "Tasmania",
"addressCountry": "Australia"
}
}
</script>
<script id="cv-inline-data">
window.__CV_DATA__ = {"metadata":{"version":"2.1.0","last_updated":"2026-03-01T00:00:00Z","data_source":"CV Pipeline","enhancement_ready":true},"personal_info":{"name":"Adrian Wedd","title":"AI Safety Researcher & Developer","location":"PO Box 331, Cygnet, Tasmania, Australia","phone":"+61407081084","email":"adrian@adrianwedd.com","website":"https://cv.adrianwedd.com","github":"https://github.com/adrianwedd","linkedin":"https://linkedin.com/in/adrianwedd","tagline":"Building the systems I study — AI safety, multi-agent frameworks, and client delivery","availability":"Currently freelancing and building — open to the right opportunity"},"professional_summary":"Systems builder, AI safety researcher, and adversarial thinker with nearly 45 years across the stack. The failure-first methodology didn't come from papers — it came from coordinating direct actions for Greenpeace's Actions unit against well-resourced opponents where getting the risk assessment wrong had real consequences. That operational security instinct runs through everything: seven years leading cybersecurity, penetration testing, and IDAM for Tasmania's public housing sector; three years of empirical AI red-teaming across 120+ models and 18,000+ adversarial prompts; and production AI systems with hallucination detection and content safety gates built in from the start. Currently freelancing across client delivery (healthcare ecosystems, multilingual sites, ops tooling) and independent AI safety research. AuDHD — the hyperfocus and pattern recognition are features, not bugs. Most interested in where complex systems break and what that reveals about how they were built.","experience":[{"position":"Freelance Developer & AI Consultant","company":"Independent","location":"Tasmania, Australia","period":"2025 - Present","type":"Freelance","description":"Client web delivery and independent AI safety research, transitioning from seven years in government. Small client base by design — deep engagement over volume. Specialising in red-teaming, adversarial testing, evaluation frameworks, and failure-first methodology for frontier AI models. Building multi-agent systems, LLM tooling, and open-source AI safety infrastructure.","achievements":["Designed and built red-teaming and adversarial testing framework for embodied AI with automated attack adapters (TAP) and cross-modal vulnerability research on VLA models","Developed evaluation frameworks with hallucination detection, content safety gates, and model monitoring for production LLM pipelines","Applied prompt engineering, RAG, and retrieval-augmented generation across multiple agentic systems","Delivered four-site healthcare digital presence with AHPRA regulatory compliance and REST API integration","Built 10+ active open-source repositories: agent orchestration, AI governance, NLP tooling, and CLI automation","Developed multi-agent orchestration system with LangGraph, LangChain, and Anthropic SDK"],"technologies":["Python","JavaScript","TypeScript","LangChain","LangGraph","Anthropic SDK","Claude API","Prompt Engineering","RAG","Red-teaming","Adversarial Testing","Evaluation Framework","Hallucination Detection","Content Safety","Multi-Agent Systems","GitHub Actions","Docker","FastAPI","NLP"]},{"position":"Systems Analyst / Acting Senior Change Analyst","company":"Homes Tasmania (formerly Department of Communities Tasmania)","location":"Tasmania, Australia","period":"2018 - Feb 2026","type":"Full-time","description":"Led systems integration, cybersecurity, and digital transformation for Tasmania's public housing sector. Delivered penetration testing, vulnerability assessment, and identity and access management (IDAM) improvements. Deployed automation and AI tooling to improve service delivery for vulnerable communities.","achievements":["Designed and implemented RESTful API integrations and SFTP data exchange for the Housing Management System","Led cybersecurity program including vulnerability assessment, penetration testing, and Essential Eight compliance uplift","Implemented identity and access management (IDAM) controls reducing unauthorised access risk","Built Python and PowerShell automation scripts for operational workflows, reducing manual processing time","Pioneered responsible AI adoption for data analysis and decision support, including prompt engineering and LLM integration","Developed ISO 27001-aligned information security practices across the department","Contributed to ICT policy development and audit processes for departmental compliance"],"technologies":["Python","PowerShell","JavaScript","RESTful APIs","SFTP","SQL","Azure","Penetration Testing","IDAM","Essential Eight","ISO 27001","Prompt Engineering"]},{"position":"ITS Client Services Officer","company":"University of Tasmania","location":"Tasmania, Australia","period":"2015 - 2018","type":"Concurrent","description":"IT support across the full university user base — researchers, academics, administrative staff. Exposure to an unusually wide range of systems and problems built the diagnostic instinct that carries through everything since.","achievements":["Sole support contact for a diverse user base with no tolerance for downtime — learned to triage fast and fix faster","Built and maintained procedural documentation that actually got used, reducing repeat escalations","Managed Microsoft 365 and Active Directory environments across a complex multi-faculty structure","Handled ServiceNow ticketing and escalation paths for infrastructure issues beyond first-line scope"],"technologies":["Microsoft 365","Windows Server","Active Directory","ITIL","ServiceNow"]},{"position":"Director","company":"Digital Agency PTY LTD","location":"Australia","period":"2015 - 2018","type":"Concurrent","description":"Ran a small digital agency focused on nonprofits and small businesses — Google Grants, AdWords, Analytics, and web presence. Learned quickly that clients don't buy strategy; they buy outcomes they can point to.","achievements":["Built and managed Google Ad Grants campaigns for nonprofits, maximising the $10k/month allowance to drive real audience growth","Set up Analytics and conversion tracking from scratch for clients with no prior measurement — made their spend legible","Delivered web and campaign work across diverse sectors: environmental orgs, health practitioners, retail","Managed client relationships end-to-end: scoping, delivery, reporting, iteration"],"technologies":["Google Analytics","Google AdWords","Bing Ads","Digital Marketing","Campaign Management"]},{"position":"Second Level IT Support Engineer","company":"The Wilderness Society Inc.","location":"Australia","period":"2012 - 2015","type":"Full-time","description":"End-to-end IT infrastructure ownership for a national environmental organisation with distributed operations across Australia. Responsible for all server environments, network infrastructure, and communication systems — the kind of breadth that builds genuine systems thinking.","achievements":["Owned lifecycle management of 60+ Windows/Debian/Ubuntu servers across a nationally distributed organisation","Maintained heterogeneous network infrastructure with no dedicated team — full-stack from physical layer to application","Built and administered Google Apps and VOIP PBX infrastructure replacing legacy systems","Executed complex migrations and decommissioning with zero data loss across diverse environments","Delivered ICT capability to teams with no technical background, building organisational resilience"],"technologies":["Windows Server","Debian/Ubuntu","Network Infrastructure","Google Apps","VOIP PBX","Server Management"]},{"position":"Actions Unit — Communications & Logistics Coordinator","company":"Greenpeace Australia Pacific","location":"Australia","period":"2010 - 2012","type":"Full-time","description":"Operational planning and execution for Greenpeace Australia Pacific's direct action campaigns. The Actions unit operates against well-resourced opponents — corporations and state actors who actively work to prevent campaign success. Work required genuine operational security, adversarial threat modeling, and risk assessment under conditions where failure meant people getting hurt or arrested.","achievements":["Planned and coordinated direct action operations — logistics, communications, and security across multiple simultaneous campaigns","Developed threat models and operational security protocols against adversarial opponents with surveillance and legal resources","Designed and maintained field ICT infrastructure for covert operations: encrypted comms, dead drops, secure channels under active counter-surveillance","Conducted pre-action risk assessments enumerating failure modes, contingency routes, and abort criteria","Trained activists in operational security and secure communications under pressure","Coordinated multi-team logistics across geographically distributed operations with real-time adaptation to adversarial conditions"],"technologies":["Operational Security","Threat Modeling","Risk Assessment","Encrypted Communications","Field ICT","Logistics Coordination","Adversarial Planning"]},{"position":"DJ / Multimedia Artist / Freelance Technologist","company":"Independent","location":"Australia","period":"~2000 - 2010","type":"Mixed","description":"A decade of freelance work across music, multimedia art, and technology. DJing, audio production, visual art, and building whatever clients or collaborators needed. The thread through all of it was systems — sound systems, visual systems, technical systems — and the instinct to build tools rather than just use them.","achievements":["Professional DJ across venues and events, managing PA systems, lighting rigs, and live audio","Created multimedia art installations combining audio, visual, and interactive elements","Freelance web development and technical consulting for small businesses and artists","Built early web presences for clients with no digital footprint, learning full-stack by necessity"],"technologies":["Audio Production","Web Development","Multimedia","Live Sound","Creative Technology"]}],"projects":[{"name":"This Wasn't in the Brochure","subtitle":"Published Book — Co-Parenting Neurodivergent Children","description":"Practical, neurodiversity-affirming guide for co-parents raising children with ADHD, Autism, PDA, and ODD. The first book written specifically for the co-parenting context. Four localised editions (AU, US, UK, NZ) with culturally-appropriate legal, medical, and educational systems. Companion children's book in development.","technologies":["Technical Writing","Neurodiversity","Research Synthesis"],"url":"https://thiswasntinthebrochure.wtf","status":"Published","period":"2025 - Present"},{"name":"NotebookLM Automation","subtitle":"CLI Tooling for Google NotebookLM","description":"CLI automation toolkit for Google NotebookLM with research quality controls, deterministic exports, template engine, JSON schema validation, and comprehensive integration testing.","technologies":["Bash","Python","CI/CD","JSON Schema"],"github":"https://github.com/adrianwedd/notebooklm-automation","status":"Active Development","period":"2025 - Present","url":"https://github.com/adrianwedd/notebooklm-automation"},{"name":"Orbitr","subtitle":"AI Music Sequencer","description":"Multi-track polyphonic AI sequencer with concentric ring interface. Inspired by Playtronica's Orbita, enhanced with Meta's MusicGen for real-time sample generation. 36 issues tracked across phased delivery.","technologies":["HTML/CSS/JS","MusicGen","Web Audio API"],"github":"https://github.com/adrianwedd/orbitr","status":"Active Development","period":"2025 - Present","url":"https://github.com/adrianwedd/orbitr"},{"name":"Evolve Evolution","subtitle":"Multi-Site Healthcare Digital Presence","description":"Four-site digital presence for a healthcare practitioner: AHPRA-compliant chiropractic clinic site, coaching/personal brand platform, author hub, and business evolution site. Shared CSS design token system, Cliniko API integration for appointment management, GA4-to-GTM analytics migration, Cloudflare Pages deployment with serverless functions, and quarterly regulatory compliance workflow. 170+ issues tracked across phased delivery.","technologies":["HTML/CSS/JS","Cloudflare Pages","Cliniko API","GA4/GTM","Structured Data"],"status":"Client Delivery","period":"2025 - Present","url":"https://evolvechiropractictas.com"},{"name":"Tanda Pizza","subtitle":"Localized Restaurant Website","description":"Static website for a Bali-based restaurant, localized to 5 languages with WhatsApp ordering integration and gluten-free menu filtering.","technologies":["HTML/CSS/JS","i18n","WhatsApp API"],"status":"Client Delivery","period":"2026","url":"https://tandapizza.com"},{"name":"Failure-First Embodied AI","subtitle":"AI Safety Red-Teaming Framework","description":"Red-teaming and benchmarking framework for embodied and agentic AI systems. Focused on adversarial testing, recursive failure analysis, recovery evaluation, and human-in-the-loop safety. Includes computer vision vulnerability research on VLA models, neural network robustness evaluation, automated attack adapters (TAP), cross-modal failure analysis, prompt injection and jailbreak testing, bias detection, interpretability analysis, model evaluation pipelines, and AI ethics documentation.","technologies":["Python","AI Safety","Red-teaming","Adversarial Testing","Computer Vision","Neural Networks","Model Evaluation","AI Ethics","Prompt Injection","Bias Detection","Robustness","Interpretability","Evaluation Framework","Content Safety","Responsible AI","Deep Learning","NotebookLM","arXiv"],"github":"https://github.com/adrianwedd/failure-first-embodied-ai","status":"Active Research","period":"2025 - Present","url":"https://github.com/adrianwedd/failure-first"},{"name":"TEL3SIS","subtitle":"Voice-First Agentic Phone System","description":"Real-time telephony platform with LLM-powered conversations, tool use (calendar, SMS, email), tri-layer memory (Redis/SQLite/vectors), safety oracle for pre-execution checks, and human handoff. Built on vocode-python with Whisper, GPT-4, and ElevenLabs.","technologies":["Python","Vocode","Whisper","Redis","Docker"],"github":"https://github.com/adrianwedd/TEL3SIS","status":"Active Development","period":"2024 - Present","url":"https://github.com/adrianwedd/tel3sis"},{"name":"Afterwords","description":"Give Claude Code a voice. Local zero-shot voice cloning TTS server for Mac — clone any voice from a 15-second YouTube clip and run it on Apple Silicon. Pairs with Claude Code for full voice conversations. No cloud API, no subscription, no data leaves the machine. ~15 voices included out of the box.","url":"https://adrianwedd.github.io/afterwords","highlights":["Zero-shot voice cloning running locally on Apple Silicon with no cloud dependencies","Complements Claude Code's /voice input with TTS output for full voice loop","Privacy-first: all processing on-device, no data leaves the machine"]},{"name":"SPARK","description":"Claude-powered robot companion for a neurodivergent child, built on SunFounder PiCar-X and Raspberry Pi 5. Voice-controlled robotics platform with three-layer cognitive architecture and REST API. Designed around the AuDHD frameworks from This Wasn't in the Brochure.","url":"https://spark.wedd.au","highlights":["Executive function scaffolding, emotional regulation, and meltdown protocol (Safety, Silence, Space)","Sideways engagement via declarative language for demand-avoidant children","Claude-powered voice loop with espeak TTS on Raspberry Pi 5"]},{"name":"Wolf Clan Zen Do Kai","description":"Public website and private operations hub for Wolf Clan Zen Do Kai martial arts club. Club information, blog with historical research, class schedules, and an authenticated hub for instructors with lesson plans, grading records, and knowledge base.","url":"https://wolfclanmartialarts.com","highlights":["Public-facing club site with SEO-optimised historical blog content","Private operations hub with instructor tools and lesson management","Built with vanilla HTML/CSS/JS, deployed on Cloudflare Pages"]}],"skills":[{"name":"Python","category":"Programming Languages","tier":"Primary","description":"AI/ML development, automation scripting, backend services"},{"name":"JavaScript","category":"Programming Languages","tier":"Primary","description":"Frontend and Node.js development, automation"},{"name":"TypeScript","category":"Programming Languages","tier":"Secondary","description":"Type-safe application development"},{"name":"PowerShell","category":"Programming Languages","tier":"Secondary","description":"Windows automation and systems administration"},{"name":"Large Language Models","category":"AI & Automation","tier":"Primary","description":"LLMs, Claude API, Anthropic SDK, OpenAI API, LangChain, LangGraph, prompt engineering, RAG, retrieval-augmented generation, fine-tuning, multi-agent systems, natural language processing (NLP)"},{"name":"AI Safety & Evaluation","category":"AI & Automation","tier":"Primary","description":"Red-teaming, red teaming, adversarial testing, evaluation frameworks, model evaluation, hallucination detection, content safety, AI ethics, responsible AI, AI alignment, failure-first methodology, FMEA, pre-mortem analysis, transformer and BERT-based evaluation, explainability, interpretability, safety testing"},{"name":"Process Automation","category":"AI & Automation","tier":"Primary","description":"Workflow automation, scripting, CI/CD pipelines"},{"name":"Systems Integration","category":"Infrastructure","tier":"Primary","description":"API development, data exchange, service integration"},{"name":"Cybersecurity","category":"Infrastructure","tier":"Primary","description":"Penetration testing (pentesting), vulnerability assessment, threat modeling, identity and access management (IDAM), Essential Eight, ISO 27001, Zero Trust, network security, information security"},{"name":"Linux/Windows Server","category":"Infrastructure","tier":"Primary","description":"Server management, networking, infrastructure"},{"name":"Docker","category":"DevOps","tier":"Secondary","description":"Containerization and development environments"},{"name":"GitHub Actions","category":"DevOps","tier":"Primary","description":"CI/CD pipeline design and automation"},{"name":"ML Libraries","category":"AI & Automation","tier":"Secondary","description":"PyTorch, TensorFlow, Hugging Face Transformers, scikit-learn, NumPy, Pandas, spaCy, NLTK — deep learning, neural networks, supervised and unsupervised learning, transformer-based evaluation, and classical machine learning, transfer learning"},{"name":"MCP Development","category":"AI & Automation","tier":"Secondary","description":"Model Context Protocol servers for Claude Code integration"},{"name":"Voice/Telephony AI","category":"AI & Automation","tier":"Secondary","description":"STT/TTS pipelines, vocode, Whisper, ElevenLabs"},{"name":"Cloudflare Pages","category":"DevOps","tier":"Secondary","description":"Static site deployment, serverless functions, email routing"},{"name":"React","category":"Frontend","tier":"Secondary","description":"Frontend application development"},{"name":"SQL / PostgreSQL","category":"Data","tier":"Primary","description":"Database design, querying, and optimization"},{"name":"Cloud Platforms","category":"Infrastructure","tier":"Primary","description":"AWS, Azure, GCP, Cloudflare — cloud deployment, serverless functions, infrastructure management"},{"name":"Kubernetes","category":"DevOps","tier":"Secondary","description":"Container orchestration, cluster management"},{"name":"Deep Learning","category":"AI & Automation","tier":"Primary","description":"Neural networks, transformer architectures, CNN, RNN, model training and evaluation, transfer learning"},{"name":"Generative AI","category":"AI & Automation","tier":"Primary","description":"Large language models (LLMs), generative model evaluation, prompt engineering, content generation pipelines"},{"name":"Computer Vision","category":"AI & Automation","tier":"Secondary","description":"VLA model evaluation, cross-modal vulnerability research, image-based adversarial testing"},{"name":"Data Engineering","category":"Data","tier":"Secondary","description":"Data pipelines, ETL, data analysis, data science, JSON/CSV processing, vector databases, Redis, MongoDB"},{"name":"MLOps","category":"DevOps","tier":"Secondary","description":"Model deployment, model monitoring, production ML pipelines, experiment tracking"},{"name":"FastAPI","category":"Programming Languages","tier":"Secondary","description":"Python web framework for API development and microservices"},{"name":"Bash","category":"Programming Languages","tier":"Secondary","description":"Shell scripting, Unix systems automation"},{"name":"Leadership & Collaboration","category":"Soft Skills","tier":"Primary","description":"Team leadership, stakeholder management, problem solving, critical thinking, cross-functional collaboration, technical presentations, agile delivery"}],"achievements":[{"icon":"📚","title":"Published Author — This Wasn't in the Brochure","description":"Wrote and published a neurodiversity-affirming guide for co-parents raising children with ADHD, Autism, PDA, and ODD. Four localised editions (AU, US, UK, NZ). Built using a three-AI collaborative workflow as an executive function accommodation — documented transparently in the book's appendix.","date":"2025","category":"Publication","url":"https://thiswasntinthebrochure.wtf"},{"icon":"🏛️","title":"Security & Systems Uplift — Homes Tasmania","description":"Led cybersecurity and systems integration for Tasmania's public housing sector: Essential Eight compliance uplift, penetration testing, IDAM controls, and RESTful API integrations for the Housing Management System. Infrastructure that vulnerable people depend on — the stakes made the work serious.","date":"2018-2024","category":"Technical Achievement"},{"icon":"🤖","title":"AI Safety Research & Red-Teaming","description":"Developed failure-first red-teaming framework for embodied AI systems with 104 tracked research issues, cross-modal vulnerability analysis on VLA models, and automated attack adapters. Pursuing commercial AI safety services.","date":"2025-2026","category":"Research"},{"icon":"📞","title":"Voice-First AI Agent Platform","description":"Built TEL3SIS, a real-time telephony platform with LLM-powered conversations, tri-layer memory architecture, safety oracle, and tool orchestration across 226 tracked issues.","date":"2024-2026","category":"Innovation"},{"icon":"🌿","title":"Technology for High-Stakes Campaigns","description":"Maintained distributed IT infrastructure across The Wilderness Society's national operations and built field technology for Greenpeace direct actions — two very different threat models, both requiring the infrastructure to be invisible until it isn't.","date":"2010-2015","category":"Social Impact"}],"education":[{"degree":"Self-Directed AI & Machine Learning Study","institution":"Independent Study","period":"2018 - Present","description":"Ongoing self-directed learning covering AI/ML, autonomous systems, and emerging technologies through courses, research, and building projects.","key_areas":["Multi-Agent Systems","Large Language Models (LLMs)","Prompt Engineering","Retrieval-Augmented Generation (RAG)","Natural Language Processing (NLP)","AI Safety & Evaluation","Red-teaming","Adversarial Testing","Responsible AI","Reinforcement Learning from Human Feedback (RLHF)"]},{"degree":"Computer Science Foundation","institution":"Self-Taught & Professional Experience","period":"2000 - 2018","description":"Self-taught through two decades of professional projects, freelance work, and relentless curiosity. No formal degree — learned by building, breaking, and rebuilding.","key_areas":["Software Engineering","System Design","Database Systems","Network Administration"]},{"degree":"First Code at Age 6","institution":"Microbee (BASIC)","period":"~1981","description":"Wrote first program at age 6 on a Microbee in BASIC — a morning routine sequencer with randomised audiovisual rewards. Built to compensate for undiagnosed ADHD executive function deficits. The instinct to build tools rather than endure problems started here and never stopped."}],"certifications":[],"languages":[{"language":"English","proficiency":"Native","level":"C2"}],"interests":["Zen Do Kai martial arts (Shodan, Sempai)","170 acres of Tasmanian bush — land stewardship and quiet","Raising neurodivergent kids (thiswasntinthebrochure.wtf)","Open source and AI safety research","DJing and audio production (former life, still wired in)","Where complex systems break and what that reveals"]};
window.__ACTIVITY_DATA__ = {"last_updated":"2026-04-26T20:13:29Z","tracker_version":"v1.5","analysis_depth":"standard","lookback_period_days":30,"summary":{"total_commits":59,"active_days":30,"net_lines_contributed":746,"tracking_status":"active","last_commit_date":"2026-04-25 21:16:22 +0000"},"data_files":{"latest_activity":"github-activity-20260426-2013.json","latest_metrics":"professional-development-20260426-2013.json","latest_trends":"activity-trends-20260426-2013.json"},"cv_integration":{"ready_for_enhancement":true,"data_freshness":"2026-04-26 20:13 UTC","next_cv_update":"Automatic via CV Enhancement Pipeline"}};
window.__AI_ENHANCEMENTS__ = {"last_updated":null,"professional_summary":{},"skills_enhancement":{},"enhancement_summary":{}};
</script>
</head>
<body>
<canvas id="curvature-bg" aria-hidden="true"></canvas>
<!-- Skip to Content (Accessibility) -->
<a href="#about" class="skip-to-content">Skip to main content</a>
<!-- Theme Toggle -->
<button id="theme-toggle" class="theme-toggle" aria-label="Toggle dark and light theme">
<span class="theme-icon">☀</span>
</button>
<!-- Main Container -->
<div class="container">
<!-- Header Section -->
<header class="header" id="header" role="banner">
<div class="header-content">
<h1 class="name">Adrian Wedd</h1>
<p class="title">AI Safety Researcher & Developer</p>
<div class="location">PO Box 331, Cygnet, Tasmania, Australia</div>
<!-- Contact Links -->
<div class="contact-links">
<a href="https://github.com/adrianwedd" target="_blank" rel="noopener noreferrer" class="contact-link github" aria-label="GitHub profile">GitHub</a>
<a href="https://linkedin.com/in/adrianwedd" target="_blank" rel="noopener noreferrer" class="contact-link linkedin" aria-label="LinkedIn profile">LinkedIn</a>
<a href="mailto:adrian@adrianwedd.com" class="contact-link email" aria-label="Send email">Email</a>
<a href="tel:+61407081084" class="contact-link phone" aria-label="Call phone">0407 081 084</a>
<a href="assets/adrian-wedd-cv.pdf" target="_blank" rel="noopener noreferrer" class="contact-link pdf" download aria-label="Download full PDF">PDF</a>
<a href="assets/adrian-wedd-cv-short.pdf" class="contact-link pdf-short" target="_blank" rel="noopener noreferrer" aria-label="Download short PDF">Short PDF</a>
<button id="print-btn" class="contact-link print-btn">Print</button>
<a href="https://adrianwedd.com/activity/" class="contact-link" aria-label="View live activity dashboard" target="_blank" rel="noopener noreferrer">Watch Me Work</a>
</div>
</div>
</header>
<!-- Main Content -->
<main class="main-content" role="main">
<!-- About Section -->
<section id="about" class="section" aria-label="About">
<div class="section-header">
<h2 class="section-title">About</h2>
</div>
<div class="section-content">
<div class="professional-summary">
<p class="summary-text" id="professional-summary">
I build AI safety tooling that turns risk concepts into repeatable tests, guardrails, and
operational monitoring. Three years empirical research on frontier AI models, focused on
red-teaming, evaluation frameworks, and failure-first methodology. Seven years translating
complex technical findings into actionable insights for government decision-makers.
</p>
</div>
<div class="core-competencies">
<h3 class="competency-title">Core Competencies</h3>
<div class="competency-grid">
<div class="competency-item">
<div class="competency-content">
<div class="competency-name">AI Safety & Evaluation</div>
<div class="competency-desc">Red-teaming, adversarial testing, failure-first methodology</div>
</div>
</div>
<div class="competency-item">
<div class="competency-content">
<div class="competency-name">Frontier AI Models</div>
<div class="competency-desc">Claude API, multi-agent systems, evaluation frameworks</div>
</div>
</div>
<div class="competency-item">
<div class="competency-content">
<div class="competency-name">Risk Assessment</div>
<div class="competency-desc">Pre-mortem analysis, FMEA, failure mode identification</div>
</div>
</div>
<div class="competency-item">
<div class="competency-content">
<div class="competency-name">Policy Translation</div>
<div class="competency-desc">Technical findings to actionable governance for decision-makers</div>
</div>
</div>
</div>
</div>
</div>
</section>
<!-- Experience Section -->
<section id="experience" class="section" aria-label="Experience">
<div class="section-header">
<h2 class="section-title">Experience</h2>
</div>
<div class="section-content">
<div class="experience-timeline" id="experience-timeline">
<div class="timeline-loading">
<div class="loading-spinner small"></div>
<span>Loading experience data...</span>
</div>
</div>
</div>
</section>
<!-- Projects Section -->
<section id="projects" class="section" aria-label="Projects">
<div class="section-header">
<h2 class="section-title">Projects</h2>
</div>
<div class="section-content">
<div class="projects-grid" id="projects-grid">
<div class="projects-loading">
<div class="loading-spinner small"></div>
<span>Loading project data...</span>
</div>
</div>
</div>
</section>
<!-- Activity Section -->
<section id="activity" class="section" aria-label="Activity">
<div class="section-header">
<h2 class="section-title">Activity</h2>
</div>
<div class="section-content">
<div class="activity-grid">
<div class="activity-panel" id="viz-streamgraph">
<div class="activity-panel-label">commits</div>
</div>
<div class="activity-panel" id="viz-languages">
<div class="activity-panel-label">languages</div>
</div>
<div class="activity-panel activity-panel--wide" id="viz-heatmap">
<div class="activity-panel-label">commits / week</div>
</div>
</div>
</div>
</section>
<!-- Skills Section -->
<section id="skills" class="section" aria-label="Skills">
<div class="section-header">
<h2 class="section-title">Skills</h2>
</div>
<div class="section-content">
<div class="skills-container" id="skills-container">
<div class="skills-loading">
<div class="loading-spinner small"></div>
<span>Loading skills data...</span>
</div>
</div>
</div>
</section>
<!-- Achievements Section -->
<section id="achievements" class="section" aria-label="Achievements">
<div class="section-header">
<h2 class="section-title">Achievements</h2>
</div>
<div class="section-content">
<div class="achievements-grid" id="achievements-grid">
<div class="achievements-loading">
<div class="loading-spinner small"></div>
<span>Loading achievements data...</span>
</div>
</div>
</div>
</section>
<section id="education" class="section" aria-label="Education">
<div class="section-header">
<h2 class="section-title">Education</h2>
</div>
<div class="section-content">
<div id="education-container"></div>
</div>
</section>
<section id="interests" class="section" aria-label="Interests">
<div class="section-header">
<h2 class="section-title">Interests</h2>
</div>
<div class="section-content">
<div id="interests-container"></div>
</div>
</section>
</main>
<!-- Footer -->
<footer class="footer" role="contentinfo">
<div class="footer-content">
<div class="footer-info">
<div class="footer-text">
<span class="footer-name">Adrian Wedd</span>
<span class="footer-separator">·</span>
<span class="footer-title">AI Safety Researcher & Developer</span>
</div>
<div class="footer-meta">
<span class="update-info">Last updated: <span id="footer-last-updated">Apr 26, 2026, 09:15 PM</span></span>
</div>
</div>
</div>
<div class="footer-bottom">
<div class="copyright">
<span>© 2026 Adrian Wedd</span>
</div>
</div>
</footer>
</div>
<!-- Scripts -->
<script src="assets/curvature-field.js"></script>
<script src="assets/curvature-init.js"></script>
<script type="module" src="assets/script.js"></script>
<script src="assets/activity-viz.js"></script>
<!-- Analytics Events (defer preserves document order: analytics-config.js runs first) -->
<script defer src="assets/analytics-events.js"></script>
</body>
</html>