AI Researcher — Internal AI Workflows & Tools

Role type: Full-time · Function: AI and technology · Location: TBD · Compensation: TBD

Role overview

The AI Researcher — Internal AI Workflows & Tools is a full-time builder role at Safina Ventures. The role exists because builds internal AI workflows, agents, automations, scripts, dashboards, and operating tools so every team moves faster with better proof. This is not a passive job description for someone waiting to be told what to do. It is a mandate for an owner who sees the system, understands the stakes, acts with judgement, and closes loops without creating more dependency around the founder or the team.

Role at a glance

• Position: AI Researcher — Internal AI Workflows & Tools, Full-time
• Function: AI and technology
• Reports to: Founder's Office
• Primary purpose: builds internal AI workflows, agents, automations, scripts, dashboards, and operating tools so every team moves faster with better proof.
• Operating principle: observe the real work, identify the bottleneck, act on the smallest useful fix, document the system, measure the result, and improve again.
• Decision standard: if the work can be decided, decide it; if it can be delegated, delegate it; if it repeats, turn it into a loop; if it is risky, escalate with facts.

Key responsibilities

Own the operating system of the role

• Maintain a clear map of the recurring work, active projects, open risks, key stakeholders, and next decisions.
• Convert ambiguous instructions into concrete outcomes with owners, deadlines, evidence, and follow-up.
• Keep the relevant Notion pages, trackers, checklists, and handovers accurate enough that another capable person can understand the state without a long meeting.
• Remove avoidable manual repetition by creating templates, automations, rituals, scripts, dashboards, or SOPs.
• Protect the team from hidden work, forgotten commitments, unclear ownership, and last-minute chaos.

Deliver the function, not just the tasks

• Understand what the function is meant to make true for Safina Ventures, then design daily actions around that outcome.
• Work with adjacent teams before dependencies become blockers.
• Raise trade-offs early, especially when time, budget, people, safety, quality, or reputation are at risk.
• Turn feedback from users, residents, guests, partners, applicants, or teammates into changes in the system.
• Report progress in plain language with numbers, links, screenshots, or examples.

Build proof and continuity

• Document what changed, why it changed, who approved it, and what should happen next.
• Keep a reliable weekly record of wins, misses, risks, and loops closed.
• Make recurring processes inspectable and teachable.
• Replace heroics with systems wherever possible.
• Leave every handover cleaner than it was received.

What ownership looks like

1. Notice — spot the friction, risk, opportunity, repetition, or unclear decision before it becomes expensive.
2. Assess — gather enough evidence to understand the real problem, not the loudest symptom.
3. Act — make the smallest responsible move that improves the situation now.
4. Coordinate — involve the right people at the right moment, with context they can act on.
5. Document — record the decision, proof, owner, deadline, and next checkpoint.
6. Close — confirm the issue is resolved, adopted, delegated, automated, or deliberately killed.

Decision rights and boundaries

Should act independently to

• Improve the working system inside the role without waiting for permission on every small decision.
• Create or refine trackers, templates, SOPs, dashboards, scripts, briefs, and checklists.
• Follow up with stakeholders and push work forward when ownership is already clear.
• Flag weak signals, incomplete information, repeated delays, and quality gaps.
• Stop doing work that has become performative, duplicated, or disconnected from the outcome, after documenting the reason.

Must escalate before

• Making commitments that create new legal, financial, hiring, partnership, safety, access, or public-facing obligations.
• Sharing private data, resident information, financial details, internal strategy, or sensitive incident context outside the approved audience.
• Changing a live system that another team depends on daily.
• Approving material spend, discounts, compensation, vendor commitments, or exception policies.
• Taking action where safety, privacy, compliance, reputation, or founder intent is genuinely unclear.

Ideal candidate

• High-agency operator who moves from ambiguity to useful output quickly.
• Builder mindset: creates systems, loops, and proof instead of only sending updates.
• Strong written communicator who can make messy context legible.
• Comfortable with Notion, AI tools, spreadsheets, trackers, and lightweight automations.
• Calm under pressure and direct about risk.
• Good judgement with people, money, time, quality, and reputation.
• Willing to do unglamorous work when the system needs it, while still improving the system so the same work does not stay manual forever.

Non-negotiables

• No hiding bad news. Reality must reach the right person early.
• No vague ownership. Every important item needs an owner and next action.
• No undocumented critical process. If it matters, it must be findable.
• No performative busyness. Output, proof, and loops matter more than hours occupied.
• No avoidable dependency on the founder for decisions that can be made responsibly within the role.

How we work — the Universal Role Standard

Every role at Safina Ventures runs on founder mindset, not employee mindset. In short, if you do the same manual work two weeks running, that is a loop you owe the company, not a routine to settle into. You are measured by loops built, hours freed, decisions improved, quality raised, risks reduced, and problems solved. Every contract runs through an 11 day non-paid trial, a 111 day probation with 11 hours notice either way, and a 1111 day long-term commitment.

Success in this role

• The function is more reliable, measurable, and easier to operate than it was before.
• Recurring work has owners, standards, and documented loops.
• Decisions happen closer to the work, with fewer unnecessary escalations.
• The team can see the current state without chasing scattered context.
• Risks are raised early and resolved with proof.
• The role creates leverage for the founder and the organisation, not more management load.

Suggested performance measures

• Loops closed per month and hours saved through those loops.
• Quality and timeliness of weekly reporting.
• Percentage of key processes documented and actively used.
• Number of unresolved items without owner, deadline, or next step.
• Stakeholder response time and follow-through reliability.
• Evidence that decisions improved because of data, documentation, or better system design.

Operating rhythm

1. Daily — inspect the live work, clear blockers, update the source of truth, and close the next obvious loop.
2. Weekly — review outcomes, misses, risks, metrics, and decisions needed from leadership.
3. Monthly — kill stale work, improve recurring systems, and identify the next leverage point.
4. Quarterly — reassess whether the function is still designed around the right outcomes.

What this role is not

• Not an assistant role that waits for instructions.
• Not a reporting role that describes problems without changing them.
• Not a specialist silo that ignores how the work affects the wider Safina Ventures.
• Not a place for hidden context, private to-do lists, or heroic last-minute saves.