Role Description We are seeking Chemists (PhD or Master's Degree) with a strong background in upper-undergraduate Chemistry or beyond to design rigorous Chemistry problems used to train and evaluate frontier AI models. This role focuses on creating realistic, exam-quality tasks that are solvable by humans yet challenging enough to expose failure modes in advanced language models. - Design and curate upper-undergraduate and graduate-level Chemistry problems with clear, fully worked solutions. - Create challenging, realistic tasks that expose failure modes in frontier AI systems. - Convert diagram-heavy or multi-part chemistry questions into self-contained, text-only problems with definitive answers. - Write rigorous explanations and verify scientific accuracy throughout the problem and solution. - Evaluate AI-generated responses, diagnose errors, and refine questions to meet difficulty targets. - Review Chemistry tasks created by others and improve clarity, correctness, and realism. - Collaborate with cross-functional teams to maintain consistent standards across large task batches. Commitments Required: 8 hours per day with an overlap of 4 hours with PST. Employment type: Contractor assignment (no medical/paid leave) Duration of contract: 2 months (possible extension) Location: Bangladesh, Brazil, Colombia, Egypt, Ghana, India, Pakistan, Indonesia, Kenya, Nigeria, Turkey, Vietnam Selection process: Home Assignment Qualifications - Master’s or Ph.D., or those currently pursuing a Ph.D./Postdoctoral degree in Chemistry, Chemical Engineering, Biochemistry, or a related field. - Proficiency in at least three of the following tools: GROMACS, LAMMPS, Psi4, PySCF, xtb, crest, Multiwfn, Cantera, CoolProp, NWChem, PyBaMM, RDKit, and Open Babel. - Strong research and analytical skills. Requirements - Proficient in at least three of the following tools: GROMACS, LAMMPS, Psi4, PySCF, xtb, crest, Multiwfn, Cantera, CoolProp, NWChem, PyBaMM, RDKit, and Open Babel.

• Investigate and categorize human‑like cognitive phenomena in language and multimodal AI models • Analyze underlying mechanisms from both computational and cognitive perspectives • Design and run experiments to evaluate model behavior, bias, and reliability • Develop concepts and prototypes to mitigate problematic behaviors in real applications • Contribute to safer, more trustworthy AI for voice assistants and infotainment systems

• Analysis of user behavior both on the functional level (e.g., usage of assistant functions) and on the interaction level (e.g., linguistic patterns, dialog structures, speech behavior) gained through qualitative and quantitative user research, experimental setups, as well as big data campaigns • Derivation of insights and user needs based following human-centered design standards • Conceptualization, design, prototyping and evaluation of system solutions that support users across the product life cycle • Potential research questions are: Learning curves, Usage behavior, Behavioral adaptation, Relationship dynamics.

Role Description Our client is three businesses working at the front edge of hospitality, design and growth in Australia. We're hiring the person who will own the AI and automation layer that sits across all three. Architecture - Design the agentic systems that run across hospitality design, hospitality strategy, and marketing workflows. - Decide the stack: which tools, which agents, which integrations, what memory architecture, what evaluation harness. - Document the architecture so it compounds rather than evaporates when you take leave. Build - Build agents using Claude, ChatGPT, Make, n8n, Zapier, the Claude Agent SDK, and adjacent platforms. - Maintain integrations across our stack: Gmail, Google Workspace, Streak, Xero, WhatsApp, ArchiCAD outputs, WordPress, Meta Ads, Google Ads, LinkedIn, Klaviyo. - Build evaluation and observability into every agent. We need to know when one fails. Marketing, Content and Growth Systems - Build AI-assisted content systems for LinkedIn, Instagram, blog, EDM and short-form video, calibrated to brand voice per client. - Architect the brand voice infrastructure: founder interviews to voice guide to enforcement layer across all content output. - Build performance reporting pipelines that pull Meta Ads, Google Ads, LinkedIn Ads, and organic data into client-ready insights. - Design ad creative generation and variation pipelines tied to performance feedback loops. - Build SEO and AI-search optimisation systems (GEO / AEO), including keyword clustering, content briefs, on-page optimisation, and schema markup. Sales, BDM and Pipeline Systems - Build lead intake systems across website, WhatsApp, referral and inbound email, with classification and routing. - Design qualification scoring against our ICP, and the agent layer that drafts initial follow-ups. - Architect the CRM rhythm in Streak: deal stage updates from meeting transcripts, proposal follow-up cadence, pipeline reporting. - Build fee proposal generation: brief in, branded proposal out, with version control and follow-up logic. - Design LinkedIn lead-scraping, enrichment, and personalised outreach workflows for BDM. - Build the trigger logic for which deals sees personally and which the system handles. Business Operations and Project Delivery Systems - Build the project workspace generation that triggers when a deal is won: folder structure, templates, access, programme, kickoff brief. - Design the reverse brief drafting agent that converts kickoff notes into structured project briefs. - Architect the cost planning workflow: from sketch to v0 cost plan in our FOH/BOH/Services/FFE/Prelims structure, with builder benchmark lookups. - Build the builder budget revision loop: receive revised budget, reconcile against cost plan, draft client memo, schedule meeting. - Design RFI log management, drawing register hygiene, and subcontractor follow-up systems. - Build the weekly client status reporting layer per active project. - Support bookkeeping rhythm with automation: Xero reconciliation triggers, invoice issuance against milestones, debtor follow-up cadence, BAS prep. Embed - Embed AI into the daily operating rhythm of three businesses. - Coach the team on how to use the systems you build. We're not buying tools, we're buying capability. Extend - Stay across emerging AI tools, agents, frameworks and platform changes. - Bring proactive ideas. Test them. Tell us what worked and what didn't. - Identify where AI extends judgement vs replaces it. Defend the line. Year One Outcomes - AI is integrated into daily workflows across all three businesses in a measurable, defensible way. - The founder recovers significant weekly hours through agent-mediated work. - Project status reporting, cost plan generation, and client comms run on assisted workflows we can prove the ROI on. - Bookkeeping rhythm (reconciliation, invoicing, debtor chase) is supported by automation. - No Filter scales content production without proportional headcount growth. - We have a documented, defensible AI operating layer that the business owns, not the vendors. Who You Are - You can architect a system and ship the first version yourself. Not one or the other. - You have hands-on experience building with LLM tooling in production. Agents, RAG, prompt engineering, evaluation, multi-step workflows. - You understand marketing, social media, content production, and growth from the inside. - You understand how sales pipelines and CRMs actually work. - You understand business operations: project delivery, process design, the rhythm of running multiple workstreams in parallel. - You're comfortable in a small business context. - You can communicate with founders, not just engineers. - You have strong opinions about tools, architecture and sequencing. - You care about hospitality, design, growth or creative work as industries. Preferred Experience - Built and deployed production agents using Claude, ChatGPT, Notion AI, Zapier, Make, n8n, or similar. - End-to-end ownership of at least one AI / automation system in a real business. - Hands-on with marketing and social tooling: Meta Ads Manager, Google Ads, LinkedIn Ads, Klaviyo, scheduling platforms, SEO tools. - Hands-on with CRM and sales tooling: Streak, HubSpot, Pipedrive or similar, plus LinkedIn Sales Nav and enrichment platforms. - Hands-on with ops and project tooling: Notion, ClickUp, Asana, Monday, Google Workspace, project management workflows. - Experience designing prompts, evaluation, memory and tool-use for agentic systems. - Exposure to interiors, architecture, hospitality, marketing, or creative services. Bonus - Experience building brand voice infrastructure or content generation pipelines at scale. - Experience with the Claude Agent SDK or equivalent agent frameworks. - A view on AI search optimisation (GEO / AEO) and how it changes brand discovery. - Performance marketing experience: media buying, attribution, creative testing, funnel optimisation. - Bookkeeping or finance operations exposure (Xero, BAS, invoicing rhythm). - A personal project, tool or workflow you're proud of and can show us. Working Style - Long-term opportunity, not a short engagement. - Strong autonomy and ownership expected from day one. - Collaborative, innovation-focused, fast-moving. - Open to experimentation, ideas, continuous improvement. - Direct working relationship with the founder. Work Arrangement - This is a 100% remote, independent contractor engagement. - To ensure alignment and transparency, successful candidates will be expected to: - Disclose any existing ongoing roles or client work. - Reflect this engagement on their LinkedIn profile (clearly marked as "Independent Contractor").