Work-package leaders for an European Commission proposal — LAIA, The Ethical Artificial Intelligence BRISECOM S.
L. (Valencia) is submitting LAIA core project as European Commission project as sole applicant: no consortium, no consortium agreement, no co-beneficiary to negotiate. Five work-package leadership positions are open. Research is contracted directly with the people who do it. Confirmation needed by 26 October. Apply here — this request is for only one question of acceptance or not per WP as director. Details at the end. ─── THE PROJECT, IN SIX LINES ─── Today's AI reasons by association. That works for prediction, and it leaves three things weak: reasoning about causes, generalising from few examples, and planning under hard constraints. A fourth problem follows — such a system cannot say why it reached a conclusion, because the explanation comes out of the same process as the answer. LAIA inverts that. The ethical and causal constraints are not filters around the model: they are objects inside the reasoning substrate, expressed as machine-checkable axioms. A neural layer proposes, a symbolic layer verifies, and anything that is not derivable is not output. Every conclusion arrives with a derivation a third party can replay — and rejection is built to the same standard as acceptance. Four years, 2027–2030. Nine work packages. What exists at the end: a published numbered axiom set in which any rule can be contested by name; an engine whose derivations can be replayed by an independent checker; a benchmark with frozen corpora and declared baselines; and one system demonstrated at TRL4 with its AI Act conformity methodology. ─── WHAT I AM ASKING ─── Five leaders, one per package. Each leader owns its deliverables and directs its own person-month budget.
• WP2
· Deep reasoning — a neural proposal layer with symbolic verification; causal reasoning with interventions and counterfactuals; derivation emission for every accepted inference.
• WP3
· Deep abstraction — concepts as compression, analogical transfer by structural alignment, generalisation on held-out domains.
• WP4
· Planning under constraints — constraint-based planning where normative and causal restrictions are part of the problem and not a post-hoc filter.
• WP5
· Certification — formal certification of the derivation format, so a third party can check a trace without trusting the system that produced it.
• WP9
· The examiner — an adversarial benchmark the system has not seen; a frozen evaluation protocol; a published verdict, including when it is negative. And, by 26 October: apply here, with one line. That is the whole requirement. When you apply, the form asks you exactly one question per Work Package (WP): Q: Do you accept WP2? A: Yes/No ... and the same for WP3, WP4, WP5, and WP9 No letterhead, no signature,



no PDF, no CV at this stage. (but can be sent) It is written down, it is dated, and it is what goes into the proposal. For anything else — a question per WP, the dossier, a conversation — message me on LinkedIn. There is where I answer. ─── WHAT I AM GIVING ───
• €9,000 per person-month — gross, employer contributions and the cost of contracting as one figure.
• 24–30 person-months per package, on your own budget to direct.
• A director's role, not an execution role: you decide how the package is done and who does it with you.
• Authorship and publication of your own results. The ethical core is published openly (CC BY-SA 4.0); the implementation under Apache 2.0. What you contribute is yours and is citable.
• Full remote. No relocation, no travel, no kick-off, no review meetings, no on-site requirement of any kind. Coordination is online and asynchronous, with one short weekly call.
• No exclusivity. You keep your post, your institution, your other projects and your own clients. This is an addition to what you already do, not a replacement.
• No employment contract — and this is deliberate: it is a services contract. You remain an independent professional, you invoice for work performed, and there is no payroll, no subordination and no employment relationship of any kind.
• Or, if your rules require it, a subcontract with your institution instead. Same total, your choice.
• A four-year programme that is funded or it is not: 48 months, one applicant, one reporting line. ─── THE ADVANTAGES, SAID PLAINLY ─── 1. It is remote and it is part-time-compatible. 24–30 person-months across four years is roughly six to eight months of full-time effort spread over 48. It fits around a job, a chair, a company or a department. 2. There is nobody between you and the work. No agency, no broker, no fee per placement, no layer of intermediaries taking a cut. The rate is the rate. 3. You are not a subcontractor of a consortium. In a normal Horizon proposal you are one partner among twelve, and the money and the credit are diluted. Here you lead a package and you sign its deliverables. 4. The governance is written and it is published. Six work families, each with a named domain and a defined decision right — formal methods, technical and AI, security, ethics and law, open science, examination and assurance. Security and assurance are not an appendix added at the end:



they are two of the six families, and one exists solely to design the examination before the system exists and to publish the verdict even when it is negative. 5. It is auditable. Person-months, timesheets, decision rights and the audit trail are in the same structure. If a line is not working it can be stopped, because there is no headcount to justify — fixed cost under €1,000 a year, money goes to work performed, not to structure. 6. The subject is worth four years. If the claim holds, an independent checker can verify a machine's reasoning without trusting the machine. That is a small number of things that would matter. ─── WHAT THIS IS NOT — read this before replying ───
• Not a job. There is no employment contract, no payroll, no social security contributions on our side, no holiday, no severance, no seniority, no benefits. If you need an employment relationship, this is not it, and I would rather say so now than three emails later.
• Not full-time and not exclusive. Nothing here prevents you from any other work, including for a competitor.
• Not funded yet. This is a proposal with a deadline of 28 October 2026 and no award decision. If it is not funded, nothing is owed by either side and no contract ever existed.
• Not a consortium. There are no partners, no work-package meetings in Brussels, no coordination paperwork to negotiate with eleven other institutions.
• Not an institute. There is no lab, no building, no equipment to inherit. What exists is a contract, a budget and a deliverable. ─── WHAT HAPPENS NEXT ─── Before 26 October Hit Apply. The form asks one question per package: Do you accept WPn? And you answer Yes or No for each one. Then message me on LinkedIn if you want the dossier, or if anything needs discussing. I send the two-page document for that package and we settle it in one call. After 26 October If the package is still open, it stays open: a work-package leader who changes between submission and grant preparation is routine. We will review the CV then, not before — before the 26th we will be only answering questions and managing applications. After the award (2027) Contract, onboarding and the full technical annex. If your institution requires the subcontract instead of a services contract, we draw it that way. -=
- Questions -=
- or if the form is not your thing: [email protected]. Same inbox, same person, and the two-page document for your package comes back from there. It is also where I answer anything that does not fit in one line — what the budget covers, what "services contract" means in practice, or whether your institution has to appear. Nothing to download, nothing to sign, no CV at this stage. #EICPathfinder #HorizonEurope #NeuroSymbolicAI #AIResearch #ResearchFunding #RemoteWork #Valencia

📌 Researcher as Institution or Individual (España)
🏢 Brisecom
📍 España