based in Sydney

I help teams build AI.
And build with AI.

I'm an independent data and AI consultant with extensive experience in the tech industry. I've advised and taught teams building AI-powered systems, including engineers from Netflix, Meta, and Amazon. I host Vanishing Gradients, where I explore cutting-edge developments in data science and artificial intelligence, and I've written for publications such as Harvard Business Review and VentureBeat.

Subscribe to Vanishing Gradients ↗ See workshops & advisory →
~/hugo/now.json
# what I'm up to
$ cat now.json
{
  "podcasts": "Vanishing Gradients · High Signal",
  "course":   "Building AI Apps · Maven",
  "writing":  "hugobowne.substack.com",
  "building": "agents · harnesses · skills",
  "based-in": "Sydney, AU"
}
$
~/hugo/stack.txt
# currently using
claude · gemini · qwen · mistral · gemma · pi
amp-code · codex · agent-skills
python · typescript · marimo · modal
pydantic · logfire · chroma · hf · obsidian
learners reached
6M+
podcast episodes
200+
newsletter subs
13K+
maven grads
350+

Podcasts

for the people building AI · and the people leading teams that build it
Vanishing Gradients
70+ eps · since 2022
Everything you wanted to know about data, ML, and AI, but were too afraid to ask. Guests include Jeremy Howard, Wes McKinney, Hamel Husain, Shreya Shankar & more.
How to Build a Coding Agent with Nico (Amp Code)with Nicolay Gerold · Amp Code
Sep '26
The Rise of the AI Scientistwith Hamel Husain
Sep '26
If Developers Build on Chinese Open-Weight Models, Who Leads AI?with Sebastian Raschka
Aug '26
Agentic engineering and the lost art of verificationwith Wes McKinney, Jeremiah Lowin & Randy Olson
May '26
LLM architecture in 2026: what you need to knowwith Sebastian Raschka · author of Build a LLM from Scratch
Apr '26
The rise of agentic searchwith Jeff Huber · Chroma
Dec '25
High Signal
by Delphina · since 2024
Deep conversations with the people shaping data science, ML & AI, for engineering leaders and practitioners.
The Verification Crisis: Why Trust Is the New Bottleneck in AIwith Noah Smith · Noahpinion
Ep 41
The Economic Reality of AI: Friction, Talent, and the Future of the Firmwith Steve Tadelis · UC Berkeley
Ep 40
The 100-year lead: what baseball teaches us about the future of AIwith Chris Fonnesbeck · PyMC, Yankees & Mets
Ep 39
The post-coding era: what happens when AI writes the system?with Nicholas Moy · Google DeepMind, ex-Windsurf
Ep 32
From context engineering to AI agent harnesseswith Lance Martin · LangChain
Ep 28

Writing

essays, opinion, and research · substack · O'Reilly · HBR
date
title
outlet
tag
2026.09.15
Beyond Navier–Stokes: Who Controls Scientific Discovery?
O'Reilly Radar
essay
→
2026.08.30
How to Build a Search Agent and Prove It Works
Vanishing Gradients
tutorial
→
2026.07.22
Stop Overengineering Your Agent Harness
O'Reilly Radar
opinion
→
2026.06.01
The Agentic Data Science Research Lab
Vanishing Gradients
essay
→
2026.04.10
AgentOps: lessons from over 1,400 production deployments of AI systems
Vanishing Gradients
essay
→
2026.03.24
How to build a general-purpose AI agent in 131 lines of Python
O'Reilly Radar
tutorial
→
2026.03.16
Build AI products locally: 10 hours of free workshops with a Google DeepMind researcher
Vanishing Gradients
tutorial
→
2026.03.11
Why most AI projects fail before they ship (and the eval-driven approach to fix it)
Vanishing Gradients
essay
→
2026.02.28
Building agents that build themselves
Vanishing Gradients
workshop
→
2025.06.27
Stop building agents
Vanishing Gradients
opinion
→
2018.08.15
What data scientists really do, according to 35 data scientists
Harvard Business Review
research
→

Workshops & advisory

cohort courses, livestreams, and advisory engagements · contact@vanishinggradients.co
01 · next cohort · oct 2026
Build Your Agentic Software Factory
Maven · with Eleanor Berger
Oct 13–22 (Sydney)
Enroll →
02 · next cohort · oct 2026
Master Agentic Data Science
PyMC Labs × Vanishing Gradients · with Thomas Wiecki & Luca Fiaschi
Oct 6–15 · Tue & Thu, 6–9 PM ET
Enroll →
03 · enquire
Private team workshops
1 to 3 day sessions on shipping LLM systems, or adopting agentic engineering.
Onsite or remote
Enquire →
04 · ongoing
Advisory & office hours
For teams and founders building AI products, or rebuilding how they ship with AI.
Monthly retainer
Enquire →

Guests, co-teachers, co-authors

a partial index · 2022–2026 · podcasts, workshops, essays
Jeremy Howardfast.ai
Fei-Fei LiStanford
Wes McKinneypandas
Hamel HusainParlance Labs
Hilary Masondata science
Michael JordanUC Berkeley
Andrew GelmanColumbia
Shreya ShankarUC Berkeley
Thomas WieckiPyMC Labs
Samuel ColvinPydantic
Cassie Kozyrkovex-Google
Travis OliphantAnaconda
Peter WangAnaconda
Katharine Jarmulprivacy
Stefan KrawczykMaven co-teacher
Matthew RocklinDask
Alan NicholRasa
Charles FryeModal
Ravin KumarDeepMind
Akshay AgrawalMarimo

Building AI Agents: the open-source course

self-paced · free · 6 sections
A self-paced course on building agent harnesses. Posts, podcasts, workshops, and code from me and my collaborators — plus links to a few external resources too. A more detailed guide here ↗

An agent is just an LLM calling tools in a loop. A harness is everything around it: tools, context, hooks, sandboxes, memory, the loop itself. The LLM is the brain. The harness is the body.
01 · start here
Build something
Give an LLM read, write, edit, and bash. You have a computer-using agent.
postHow to build a general-purpose AI agent in 131 lines of Pythonsubstack · the same 4-step pattern for code & search agents
externalSame piece on O'Reilly Radaroreilly.com
codehugobowne/building-with-aigithub · runnable agent
02 · next
Agents that build themselves
Hot-reloading tools, markdown memory, hooks, sandboxes. With Ivan Leo.
postBuilding Agents That Build Themselveswith Ivan Leo · Google DeepMind, ex-Manus
workshopLivestream: the 2-hour live buildyoutube
workshopBuild Your Own Deep Research AgentGemini, Exa, parallel subagents, cited reports
codebuild-your-own-ai-assistantgithub
03 · deepen
Harness engineering
The deeper conversations. Why context engineering is the new system design.
podcastWhy Agent Context Isn't Enoughwith Jeff Huber · Chroma
podcast3 Principles for Context Engineeringwith Lance Martin · Anthropic · Reduce, Offload, Isolate
podcast15 Privacy Questions Every AI Builder Is Askingwith Katharine Jarmul · jump to Q4
04 · specialize
Agentic search
Where a lot of this is heading. The traditional search stack falls away.
workshopHow to Build a Search Agent and Prove It Workswith Doug Turnbull & John Berryman
codeWorkshop code and evalsBM25 baseline, agent loop, and retrieval evals
postHow to Build Your First Agentic Search Applicationwith Doug Turnbull · +15-30% NDCG, no training
podcastEp 68: A Builder's Guide to Agentic Search & Retrievalwith Doug Turnbull & John Berryman
05 · evaluate
Evals: please do them
End-to-end task success first, then step-level diagnostics.
postHow Evals Are Central To Harness EngineeringAntaripa Saha · traces, failure analysis, and harness iteration
podcastThe Rise of the AI Scientistwith Hamel Husain · verification and product design
externalEvals FAQ: How do I evaluate agentic workflows?Hamel Husain · transition failure matrices
externalDemystifying Evals for AI AgentsAnthropic · tasks, trials, graders, multi-turn
Want it taught live? I also run team workshops and cohort courses on building agents, and on building with AI. Email contact@vanishinggradients.co to find out more.

subscribe_to(vanishing_gradients)

Essays, a new podcast, and occasional workshop invitations, in your inbox every week. Read by 13,000+ practitioners and leaders.

13K+ readers · Substack · no spam, ever
subscribe on substack ↗ read the archive →