×
Metacognition
A persistent-memory and reasoning layer that sits underneath LLMs
Last verified September 11, 2026 · Updated daily
What Metacognition is building
Metacognition is building the layer that sits *underneath* a large language model rather than on top of it. The company's argument is that today's models are text predictors with no state: they are briefed fresh every session, forget what they learned, and cannot be inspected afterwards. Metacognition's architecture separates language from knowledge representation, so the model does the talking while an explicit, persistent knowledge store does the remembering. Around that sit three capabilities the founders keep naming: memory that survives across sessions, reasoning that adapts from prior experience, and controls over what actions an agent is actually permitted to take. The first product is Zeus, which the company calls an operating system for agents. Zeus governs how models interact with machines and external systems, which is what turns a chat interface into something that can be handed a physical process. The team's own shorthand for it is 'Windows for robots': the missing software platform that lets robot hardware be trained to do useful tasks by non-experts giving simple verbal instructions and verbal feedback, rather than by an engineer writing task-specific code. The initial targets are industrial, not consumer. The company has demonstrated simulations for directing open-cut mining operations and for managing electricity grids through voice-based commands. Both are cases where the useful thing is not fluent language but an agent that accumulates context over months and whose decisions can be inspected afterwards. Safety is architectural here rather than bolted on. The company has adopted Australia's voluntary AI Safety Standard as a baseline and says it has additional internal benchmarks built into the architecture from inception, with van den Hengel stating they are 'prioritising building safety and human control into the design.'
Why this matters
The credentials behind this are unusually heavy for a pre-seed. Anton van den Hengel has an H-index of 92 across 400+ papers and 40,000+ citations, grew the Australian Institute for Machine Learning to 130 researchers, and spent four years as a Director of Machine Learning at Amazon. Stephen Gould has an H-index of 63 and 27,000+ citations, was a Principal Research Scientist at Amazon, and co-founded Sensory Networks, which Intel bought in 2013. This is not a team learning the field on the round. The thesis is also a real gap rather than a positioning exercise. Gould's framing is the sharpest version of it: 'Right now, using AI feels a bit like working with someone who's briefed fresh every meeting.' Agent memory today is mostly retrieval bolted onto a stateless model, and it degrades exactly where industrial deployments need it most — over long horizons, across sessions, under audit. Whether an explicit knowledge-representation layer beats bigger context windows is genuinely unsettled, which is the point: Main Sequence is underwriting a research bet, not a go-to-market one.
Open roles at Metacognition
3 positions we're tracking. Roles are re-checked daily and removed when filled.
Senior Research Scientist (remote/flexible, reports to the Chief Scientist)
First seen today
Senior Machine Learning Engineer (flexible location, reports to the Chief Engineer)
First seen today
Data Scientist
First seen today
Know when Metacognition is hiring before anyone else
A role stays uncontested for about four days. Here's the window — and where we put you in it.
From $9/month, cancel any time.
Watching Metacognition
0 applicantsRole spotted & verified
1You get the alert
1You've applied
~8Hits the job boards
250+Hiring outlook
Three research and engineering roles are live on the company's own site right now, and the company has publicly stated it intends to scale from roughly 10 employees to 100 within the next year. It is also explicitly recruiting internationally, not just in Adelaide.
Working at Metacognition
Metacognition: A persistent-memory and reasoning layer that sits underneath LLMs. Founded 2026, currently <50 employees. For a AI company this size, the reality is broad remit, direct access to founders, and equity that still means something if the company works out.
Most Metacognition jobs are based in Adelaide, Australia.
How to actually get hired at Metacognition
Why applying the normal way doesn't work
At this size (<50 people), Metacognition has no recruiting function. Founders handle hiring alongside everything else. Reach them directly or get lost in the inbox.
Who to contact at Metacognition
Founding team
Professor of Computer Science at the Australian National University, ARC Future Fellow and Amazon Scholar. Before academia he co-founded Sensory Networks, which Intel acquired in 2013, and he later served as a Principal Research Scientist at Amazon. He holds degrees from the University of Sydney and Stanford, has an H-index of 63 with 27,000+ citations, and teaches graduate courses on deep learning and differentiable optimization.
The commercialisation half of the founding team. Dalby is a research strategist based in Adelaide with roughly 20 years of experience turning research programs into funded ventures, having secured more than $600M in funding over his career. He is affiliated with the University of Adelaide and works across business development, innovation and coaching strategy.
Professor of Machine Learning at the University of Adelaide, Chief Scientist at the Australian Institute for Machine Learning and Director of the Centre for Augmented Reasoning. He built AIML into Australia's largest university-based machine learning research group, growing it to 130 researchers, and spent four years as a Director of Machine Learning at Amazon. He has an H-index of 92 across more than 400 papers and 40,000+ citations, and won the CVPR Best Paper Prize in 2010.
What to show them
A short, self-contained write-up of one continual-learning or memory experiment you ran yourself, with numbers. Pick a setting where a stateless model with a long context loses to an explicit memory or knowledge-representation approach, and show the forgetting curve over sessions. Two pages and a repo beats a CV. If you have a NeurIPS/ICML/ICLR/CVPR/ECCV paper in reasoning, memory, multimodal learning or large-scale training, lead with the one result in it that you would now do differently.
A cold email that works at Metacognition
What Metacognition screens for
Main Sequence is CSIRO-backed and underwrites science risk, not distribution risk — which is why a 2026-founded company with ~10 people and no revenue story raised A$10M at pre-seed. That shapes the candidate's edge: the Senior Research Scientist role explicitly includes publishing at top venues and mentoring PhD interns, so this is one of the rare startup jobs where an academic track record converts directly instead of being discounted. The hiring window is narrow and obvious: a company going from 10 to 100 in a year makes its research hires first and its process later. The Senior ML Engineer role is the quieter opportunity — it is the bridge job, taking experimental code and making it reliable and scalable, and far fewer people apply to those than to the scientist posting.
Don't send a generic CV to Metacognition. Mirror the job posting's language to get past automated screening.
Don't make these mistakes
Do not open with enthusiasm about agents or memory as a category. Anton and Stephen have spent decades publishing in exactly this area and have roughly 67,000 citations between them; 'I believe agents need long-term memory' is a sentence they have read a thousand times. Equally, do not send a portfolio of LLM wrapper projects or benchmark scores on a public leaderboard. The Senior Research Scientist posting asks for a publication record at NeurIPS/ICML/ICLR/CVPR/ECCV and demonstrated *independent* research direction — so lead with a specific result: a continual-learning setup where you measured catastrophic forgetting, or a case where explicit knowledge representation beat a longer context window, and what the numbers were.
Mistakes that kill Metacognition applications
A recycled CV gets rejected fast at Metacognition (<50 people). They notice.
Lead with their problem, not your ambition. Metacognition is focused on The credentials behind this are unusually heavy for a pre-seed — show you understand that.
Don't apply and wait. The median AI application gets no response ever. One follow-up at day five roughly doubles reply rates.
Applying to Metacognition? Get the contact, not the form.
The Metacognition interview process
3 stages · 7 days typical · take-home: yes · modelled from similar companies
We don't yet have verified candidate reports for Metacognition. What follows is the typical process for a <50-person AI company — treat it as a model, not confirmed detail.
Interview stages
Intro Call
Video call · 30 min
Culture fit and role expectations
Founder or hiring manager
Technical Deep Dive
Video call or in-person · 60 min
Past projects and problem-solving approach
Technical founder or lead
Final Round
In-person or video · 45 min
Team fit and offer discussion
Founding team
Metacognition take-home assignment
Metacognition includes a take-home exercise in their interview process. For AI roles, this typically involves a practical problem that takes 2-4 hours. Focus on clean, working code over premature optimization. They're evaluating how you think and communicate, not just the solution.
Metacognition interview timeline
At days, Metacognition's process is faster than typical for AI (10 days at this size).
Interviewed at Metacognition?
Tell us how it went — stages, questions, timeline. Takes 90 seconds and it's how this page stays accurate for the next person.
Submit your Metacognition interview experience →Metacognition jobs, frequently asked questions
How many jobs does Metacognition have open?
Metacognition currently has 3 open roles, last verified September 2026.
Does Metacognition hire remotely?
Currently, Metacognition only has in-office roles in Adelaide, Australia.
What roles is Metacognition hiring for?
Metacognition is hiring across Data, Engineering. The most recent opening is Senior Research Scientist (remote/flexible, reports to the Chief Scientist).
How do I apply for a job at Metacognition?
Apply directly through the links above, or read our guide on how to actually get hired at Metacognition.
Does Metacognition respond to cold emails?
We're still collecting cold email data for Metacognition.
Who is the hiring manager at Metacognition?
At this size, hiring is usually run by a founder or department head.
How competitive is it to get hired at Metacognition?
Roles at <50-person AI companies typically draw 50-100 applicants in the first two weeks. Applying inside 72 hours of a posting going live is the single biggest lever you control.
How many rounds is the Metacognition interview?
3 stages: Intro Call, Technical Deep Dive, Final Round.
Is the Metacognition interview hard?
It concentrates on technical depth and system design rather than abstract puzzles. The stage candidates find hardest is Technical Interview.
Does Metacognition give a take-home task?
Yes, Metacognition includes a take-home assignment.
How long does Metacognition take to get back to you?
Around 7 days across the full process.
What should I prepare for the Metacognition interview?
Study technical depth and system design. At this size (<50), they care about self-sufficiency over textbook knowledge.
Where is Metacognition based?
Metacognition is headquartered in Adelaide, Australia.
Get Metacognition roles before they're posted
A role stays uncontested for about four days. Here's the window — and where we put you in it.
From $9/month, cancel any time.
Watching Metacognition
0 applicantsRole spotted & verified
1You get the alert
1You've applied
~8Hits the job boards
250+