# Alex Revill: Professional Profile ## Meta - **Canonical URL:** https://alexrevill.com - **Last Updated:** 2026-06-08 - **Version:** 0.3.0 - **Purpose:** Canonical, machine-readable source of truth for Alex's professional experience. Targeted CVs and applications are generated/curated from this store. --- ## Core Identity **Name:** Alexander Revill **Professional Positioning:** Scientist-turned-operator working at the frontier of AI and emerging capabilities. Built and ran innovation, horizon-scanning, and stage-gated pipeline functions inside a large R&D organisation (GSK), founded early-stage ventures, and now designs and builds AI-native tools and systems hands-on. **Current Location:** Sydney, Australia (relocated from London in 2025). **Contact:** hello@alexrevill.com **External Profiles:** - LinkedIn: linkedin.com/in/alexrevill --- ## Experience ### ProblemKit | Founder | March 2024 – March 2026 (discontinued) **Type:** Founding · **Relationship:** Founder · **Commitment:** Full-time · **Compensation:** Self-funded **Status:** Discontinued. Also served as the home entity under which several venture experiments ran; those are captured under Projects. #### Context Most entrepreneurship advice assumes you already have a problem worth solving. But problem *selection* itself is underexplored — founders chase solutions looking for problems, or stumble into problem spaces they're poorly positioned to serve. I started ProblemKit to develop and commercialise methodology for this pre-ideation phase: helping founders find, characterise, and validate problems worth solving rather than hunting for business ideas. #### What I Did Designed a structured methodology for autoethnographic problem discovery, characterisation, and validation, and commercialised it across several channels: **The Great Problems Logbook (self-published, hands-on):** A physical workbook containing the methodology. I wrote all the content myself — LLMs weren't yet good enough to write it convincingly — and produced the book resourcefully: I found a corporate-merchandise publisher whose customisation portal was just sufficient to produce a rudimentary book, which I worked out was a far cheaper and more instructive way to test demand than any alternative. I set up a Shopify store and did the marketing and shipping myself from home. Shipped roughly 150 copies to founders in the UK, US, and Europe (a mix of online orders and manual sales at events). I'd always wanted hands-on experience shipping a product from my living room. I tried paid ads but CAC was too high for such a competitive content field, and I couldn't see a route to bring it down — so most sales came through talks I gave and my organic LinkedIn presence. **Speaking and Workshops:** Talks and workshops at the London Longevity Hackathon, African Blockchain Challenge, MassChallenge Accelerator, Ventures Podcast, and PLUGGED co-working space. Formats from 20-minute talks to 2-hour interactive workshops, reaching hundreds of founders. **Private Advisory:** Worked 1:1 with early-stage founders on problem validation — stress-testing assumptions and sharpening problem statements before they built. **Venture Experiments:** As part of methodology development, I ran my own validation experiments across deliberately diverse problem spaces — stress-testing the methodology while exploring genuine opportunities. One explored fungal root inoculants (a product to stabilise exotic plant growth by replicating the microbiological environment of plants' native ecologies); I validated the concept with experts at Kew Gardens, then killed it on discovering it would require multi-year grant-funded research to establish the scientific basis. The opportunity was real for commercial nurseries and retailers; the commitment exceeded what made sense for a methodology experiment. Another tested a "lighter" hypothesis with performance sweatbands; after diving deep on jacquard weaves and French looms, I recognised I had no genuine passion for garment design or technical-fabric problems — validating that founder–problem fit matters as much as market fit. Both fed back into the methodology as live demonstrations of disciplined kill decisions. #### Skills Demonstrated Methodology design and productisation · resourceful solo product development and commercialisation · workshop facilitation and public speaking · disciplined kill decisions. #### Verification - The Great Problems Logbook: ~150 copies shipped — verifiable (Shopify store now closed; order/shipment records available on request). - Speaking engagements — verifiable via event organisers. --- ### FANKS | Founder | November 2025 – 27 February 2026 (discontinued) **Type:** Founding · **Relationship:** Founder · **Commitment:** Full-time · **Compensation:** Pre-revenue **Status:** Discontinued (27 February 2026). Killed on commercial judgment — the brand/supply side wasn't structurally ready, for reasons outside the product's control (below). #### Context Running community events, I saw the challenge of sustainable community monetisation firsthand. Brand partnerships are one obvious revenue model for IRL communities, but good matching infrastructure doesn't exist — brands can't easily discover and evaluate grassroots communities at any meaningful scale, and community owners lack tools to demonstrate their value and easily implement offers with sufficient data to present back to brands. This led to FANKS. #### What I Did Built infrastructure for brands to run scaled field-marketing campaigns through IRL communities (sport & fitness, entertainment, professional groups), who could in turn showcase their communities and deliver tracked offers to members. **MVP development (hands-on AI build):** Vibe-coded the MVP end-to-end on top of Luma, reading from Luma's API and triggering automations around the event flow. The MVP issued attendance rewards with tracked claim rates linked to verified attendance — proof for a later model that might automate supply of digital and physical goods (freebies, partner discounts) with proper stock handling and reporting back to brands. **Real-world test:** Tested the MVP at one bouldering event (15 attendees). I negotiated premium electrolytes from a popular D2C brand, essentially at-cost, and distributed them with proper claim tracking in place. Functionally, it worked. **Community traction:** Signed 15 community owners (pre-revenue), representing ~1,000 regularly-meeting members across London. Easy sign-ups, partly because — unlike competitors — we didn't require owners to maintain a separate community profile on a marketplace. **Competitor mapping:** Fully mapped the competition and their business models. Some leaned too hard into sample-delivery logistics and became hamstrung as small agencies; others targeted the brand side first but solved no real problem for community owners (often requiring those separate profiles), stymying growth. We deliberately differentiated. **Disciplined kill:** The kill signal was multi-front. The set of brands thinking progressively about IRL community marketing is surprisingly small and concentrated (especially in fitness and tech/networking). Many brands that lean into IRL community simply run their *own* communities rather than sponsoring independent ones, because premium brands are protective of presentation context and won't cede it. The result, where they do sponsor, is events that feel transactional — people attend for freebies, not the experience. Conversely, independent communities are increasingly protective of their experience and resist anything that feels like selling — which is itself in tension with the fact that the friction keeping communities special also caps their commercial scale. I found almost no one who'd struck this balance on either side. My conclusion: the brand side would need to be much larger and more competitive before enough brands would cede control on terms that keep communities authentic. Until then, the "run campaigns at global scale" part of the proposition falls flat. The problem is real; the solution we reached for was blocked by a downstream problem outside our product's control. I updated my thesis, killed the product, and moved on. #### Skills Demonstrated Venture validation (problem discovery, user research, dog-fooding) · hands-on (vibe-coded) MVP development on a third-party API · competitor/business-model analysis · community-owner sales and onboarding · commercial judgment (disciplined kill with a named, reasoned signal). #### Verification - Build exists, with a real Luma-calendar API integration, community-owner onboarding, a reward-flow builder, and a token-based claim/check-in flow — verifiable (local repo). - Community sign-ups, event-test numbers, and the brand partnership — self-reported (internal records). --- ### Agent-Native CV / Canonical Store (this project) | January 2025 – present (living) **Type:** Personal infrastructure / AI-native system · **Relationship:** Creator · **Commitment:** Ongoing · **Compensation:** None (personal asset) **Status:** Living. Serves as the canonical store from which I generate applications and other content. Build date: January 2025. #### Context Two things prompted this. First, people increasingly use AI to *write* compressed PDF CVs that AI-powered ATS systems then *read* — which is backwards. Second, platforms like Jack & Jill and Hatch emerged as LinkedIn alternatives, using AI to capture richer candidate input and build expanded profiles to guide career search. I liked the concept, but not that these platforms own the canonical source *and* serve the employer side of the market — a structural trust problem over time. So I built a profile where I own my architecture and data. #### What I Built A machine-readable canonical store of my professional experience: a schema.org JSON-LD backbone, narrative write-ups, an llms.txt entry point, and a per-claim verification system (verified / verifiable / self-reported, each with a method). It's designed to be queried by AI agents rather than scanned by humans, and it's now the single source of truth for my professional profile, with far more information than my LinkedIn or any two-page CV. The defining architectural choice is strategic, not technical: **users interact with my data using their own agents, not mine.** That avoids the bias inherent in platforms where the platform's agent interprets profiles on behalf of paying customers. The store provides data; the reader's own assistant reasons over it. A recent re-architecture sharpened this. The store now runs a private source layer with a curated public build and a lint guardrail that prevents private material (e.g. named references) from ever shipping. I also made a deliberate decision **not** to add an MCP server: an MCP integration would reintroduce the very friction the thesis exists to remove — the whole point is that anyone can just paste the URL and ask their assistant, with no setup. Choosing not to build the more "impressive" technical feature, because it would undermine the proposition, is the judgment I want this project to demonstrate. #### Skills Demonstrated AI-native information architecture · per-claim verification system design · agentic-web fluency · thesis development and articulation · forward product judgment (including knowing what *not* to build) · shipping tangible artefacts to demonstrate ideas. #### Verification - You're reading it — verified. --- ### Zinc VC | Entrepreneur in Residence / Fellow | October 2023 – October 2025 **Type:** Fellowship / Incubator · **Relationship:** Entrepreneur in Residence; Fellow · **Commitment:** Full-time (6 months), then episodic advisory · **Compensation:** Stipend (EIR), unpaid (Fellow) #### Context After six years at GSK, I wanted to develop my venture-building skills outside a corporate environment — harsher constraints on capital and resources, direct accountability to users. Zinc's health-focused incubator provided the structure while I explored the fertility/family-planning space, a problem I gravitated to after recognising with my co-founder that, like many peers with busy city lives and high costs, I was delaying starting a family and anxious about it. #### What I Did **Fawn Technologies (Co-Founder, October 2023 – April 2024):** Co-founded a startup addressing the public–private gap in preconception health, helping busy couples in expensive cities start healthy families with lower fertility and pregnancy risk. In roughly four months on a shoestring: - Conducted 70+ deep user interviews within three weeks (personal networking + word of mouth) and root-cause analysis on the public/private gap. - Users voluntarily indicated willingness to pay up to £3,000 for conception-support services. - Self-taught Meta advertising and ran a fake-door campaign on a £100 budget to demand-test fertility forecasting. - Achieved 2–4× industry-benchmark CTR with 20% landing-page sign-up conversion. - Proposition: "Fertility for people who don't want kids (yet)" — disproving the assumption that people don't think about family planning until actively trying. - Built business model, pricing, and GTM against a £100Bn+ TAM; assembled a clinical + data-science advisory board from senior industry roles (incl. King's College London and Imperial College London). Finding a co-founder and ideating a viable venture within two months is fast. We recognised, toward the end of the programme, that our visions for the company weren't aligned enough to accept investor money, so we wound down cleanly *before* demo day — a clean decision made early, before external capital was at stake. **Zinc Fellow (July 2024 – October 2025):** Invited into Zinc's Fellow network to advise subsequent cohorts. Advised 20+ founders across digital health, neuroscience, commercial validation, and founder–problem fit — consistently requested from a pool of 200+ Fellows. #### Skills Demonstrated Full-stack early venture validation (problem discovery, user research, quantitative testing, business model and GTM design) · self-directed learning (Meta ads, landing-page optimisation) · advisor relationship-building · knowing when to stop. #### Verification - Zinc programme participation & Fellow membership — verifiable (programme records). - Fawn validation metrics — self-reported (internal records / Meta Ads Manager). --- ### GSK | Senior Manager, Strategy — Digital Health & Innovation | March 2020 – October 2023 **Type:** Employment · **Relationship:** Employee · **Commitment:** Full-time · **Compensation:** Salaried #### Context Promoted from Technology Associate to Senior Manager — GSK's fastest-promoted graduate at the time. The team evolved from an innovation unit into owning GSK's global digital-health strategy. My role shifted toward strategic leadership: building the methodological infrastructure for the innovation pipeline and serving as the primary voice translating technology trends for senior leadership. #### What I Did **Digital Trends intelligence (C-suite reporting):** Owned and authored the "Digital Trends in Healthcare" monthly memo for three consecutive years — the primary channel through which GSK's senior leadership (CEO, CDTO, and 200+ global leaders across business and technology) tracked emerging technology relevant to pharma strategy. Authored 50+ monthly C-suite intelligence briefs, influencing $3m+ in digital-health investment decisions. Scope spanned additive manufacturing, health-data interoperability, quantum computing, in-silico drug discovery, decentralised clinical trials, synthetic data, digital therapeutics, large language models, and more. **Generative-AI early warning:** Reported on generative AI from GPT-1 through GPT-3. In 2022 I published an experimental memo containing LLM-written articles advocating safe internal experimentation. Within 48 hours, it triggered a spike in internal ChatGPT usage and a company-wide lockdown on GPT usage; the two-week pause produced a formal data-privacy and security policy enabling governed exploration. (This is the flagship AI-currency proof point — see Projects.) **Innovation methodology design:** The initial team was largely populated with people who already had ideas they wanted to build. After that pipeline was exhausted, we needed a way to reliably initiate strategically-aligned workstreams from market signals and trends. I built and delivered the methodological infrastructure that solved the team's founding pipeline problem. - *Scenario method:* a structured approach for extracting technology-trend implications from domain experts — scientists and engineers wrote two-page fictional mini-stories about the future of a domain, reviewed in film-industry-style "table reads". Ran 30+ scenario reviews, generating the bulk of the internal pipeline; later taught to three additional teams across Development, Regulatory, and Commercial. - *Stage-gated pipeline:* discovery → development → alpha → beta phases, enabling leaders to allocate resources on demonstrated traction. 50+ projects entered the pipeline, driven by 20+ technologists. - *Design-thinking and product discipline:* all internal builds were developed within a rigorous design-thinking framework. **Inflammatory-monitoring venture:** Owned the operational lead role on a flagship digital-health venture concept, in partnership with Accenture's venture-building practice. In public-safe terms: devising a novel consumer proposition that enabled monitoring and modelling of systemic inflammation relevant to the speciality-medicines portfolio and the company's broader immune-science positioning. Integrated desirability/feasibility/viability analysis for the investment case. Deprioritised following GSK's pivot to pure-play pharma (Consumer Health divestiture), but advanced to investment-ready stage with usability studies, feasibility analysis, and qual+quant demand validation complete. **International workshop delivery:** Designed and delivered innovation workshops to senior leadership across US Commercial, UK Development (Stevenage), China, Singapore (APAC), and Japan Commercial — 15–30 participants each, SVP/regional-president sponsorship. #### Outcomes 50+ strategic projects seeded · 200+ senior leaders receiving monthly intelligence · company-wide AI policy catalysed by early LLM reporting · three teams trained on scenario method · improved Development↔Commercial collaboration. #### Skills Demonstrated Strategic technology analysis and translation · methodology design and systematisation · C-suite communication · workshop design and facilitation · venture development (0→1) · cross-functional stakeholder management. #### Verification - Employment — verifiable (LinkedIn bidirectional; references available on request, held in the private layer only). - 50+ briefs / $3m+ influence / pipeline metrics — self-reported (records no longer accessible). --- ### GSK | Technology Manager, Beyond Innovation | September 2018 – March 2020 **Type:** Employment · **Relationship:** Employee (Graduate Programme → Permanent) · **Commitment:** Full-time · **Compensation:** Salaried #### Context Joined the Beyond Innovation team at its inception, having negotiated a transfer from the ERP function mid-graduate-programme — I pitched the CDTO directly in an elevator conversation, explaining my computational-neuroscience background and the team's interest, and she personally intervened to enable the move. Subsequently fast-tracked to a permanent role earlier than any other graduate company-wide. #### What I Did - Founded and authored the Digital Trends in Healthcare memo (later scaled in the Senior Manager role). - Seeded new digital ventures on behalf of the CDTO; developed the initial portfolio of strategic themes and scenario-based forecasting. - Designed and implemented the team's innovation methodology, value framework, metrics, governance, and pitch templates. - Built compounding market-intelligence infrastructure used across the team — an early, pre-Heckle ancestor of the "compounding org-memory" idea I now work on. - Designed and delivered workshops to leadership teams across the US, UK, China, Singapore, and Japan. #### Skills Demonstrated Initiative and agency in career navigation · strategic analysis and forecasting · methodology design · cross-functional workshop delivery · senior stakeholder management. #### Verification - Employment — verifiable (LinkedIn bidirectional). - Graduate fast-track timing — verifiable (HR records). --- ### GSK | Technology Associate, Global Applications & Development Tech | September 2017 – September 2018 **Type:** Employment · **Relationship:** Employee (Graduate Programme) · **Commitment:** Full-time · **Compensation:** Salaried #### Context Joined GSK's Business & Technology Consulting graduate programme in the Global ERP function, working alongside experienced SAP professionals on GSK's decade-long global template implementation — operating within large-scale, high-stakes infrastructure with zero tolerance for production interruption. #### What I Did - Contributed to SAP Direct Procurement template deployment at two of GSK's largest manufacturing sites (US and UK) with uninterrupted production. - Led a template change to the materials-management system. - Delivered end-user training; tested and upgraded functionality in a highly regulated environment. - Served as Tech Liaison for the graduate programme for a year (cross-programme feedback; networking for 50+ peers). - Earned TERP-10 foundational SAP certification; passed first time. #### Skills Demonstrated Large-scale enterprise systems · regulated-environment operations · stakeholder training and communication. #### Verification - Employment — verifiable (LinkedIn bidirectional). - TERP-10 — verifiable (certification records). --- ## Out Venturing A personal community project: an IRL event series bringing founders and curious people together through shared recreational activity (bouldering and similar). I run it because I value building community through direct, in-person connection rather than online. It has useful ancillary benefits — a real-world network in the founder scene — but it's a standalone personal endeavour, not a commercial venture. 92 subscribers on the Luma calendar; five events run to date. **Verification:** Event history verifiable via lu.ma/outventuring. --- ## Projects > A general-purpose collection of standalone projects. It houses Alex's AI / emerging-tech work (the strategic core for an AI Innovation & Emerging Capabilities remit) but isn't AI-only — AI-ness is carried by tags, not the container. Each follows a "graceful account": the bet, what was built, the role, the tech, the honest scale, and the outcome or judgment. Experimentation **is** the evidence — build fluency plus forward judgment (including disciplined kills). Gaps are marked, not invented. ### Heckle | Active (informal / pre-launch, experimental side project) - **The bet:** IRL pitch forums are an underexploited edge for founders in a noisy world — and "pitchware" can lift engagement and conversion from every audience while giving richer feedback to hosts and presenters than tools like Slido. (Insight came from attending and speaking at hackathons, demo days, and founder socials — Startmate's Sunrise demo day and an Antler pitch night specifically — where I saw founders bolting on broken QR codes and CTAs, and audiences with feedback but no private channel to give it.) - **What I built:** A functional MVP — presenters cast content through a Heckle screen the host controls; the audience joins once and has each presenter's canonical links (socials, landing page) one tap away throughout; complete post-event summaries and recaps go back to every attendee. I built the app, the site, chose the name, and worked out a monetisation model to sense-check viability. I built it functional enough to demo to a software-engineer friend and pitch him the concept; he's now taking over the technical build. - **Compounding org-memory (the AI-native architecture):** Separately, I'm experimenting with the "AI-native startup" layer around it — an exploration of baking AI into a company's foundational operations to realise compounding gains, and an evolution of the "second brain" idea into a knowledge base that self-maintains using agents. Concretely: a private, version-controlled GitHub repo organised on PARA (projects / areas / resources / archives) with a raw inbox. Raw, immutable source material (meeting transcripts captured via Granola, founder chat via a Telegram listener bot) is kept strictly separate from the synthesised PARA layer, which coding agents (Claude, Codex) file and maintain by reading an instruction set — plus a scheduled "janitor" maintenance agent (GitHub Actions) for upkeep. Live task state stays in Linear and large media in object storage (Cloudflare R2), linked rather than copied in. It is deliberately not an embeddings/vector-store RAG system — it's agent-maintained plain markdown over git, which keeps it transparent, diffable, and owned. A decision made in chat can be logged as retrievable, actionable context, reducing the documents we maintain by hand. Honest current state: scaffolded and structured; the capture pipelines and scheduled janitor are wired/stubbed and still being brought fully online. - **Role:** Originated the insight, built the MVP and site, named it, modelled monetisation, planned GTM, and designed the org-memory architecture; handing the technical build to a software-engineer collaborator. Evidence I can take technical work far enough to communicate a concept for an engineer to take over. - **Tech (verified from the repos):** *Heckle app* — Next.js + TypeScript + Tailwind, Supabase (Postgres / Auth / Realtime), Drizzle ORM, WebRTC peer-to-peer with Cloudflare Calls for TURN, Resend, QR join. *Heckle site* — Next.js + React + Tailwind with shadcn/Radix UI, Supabase, deployed on Vercel. *Org-memory* — git/GitHub, PARA markdown, Granola + Telegram capture, GitHub Actions janitor agent, Linear, Cloudflare R2. All assembled via Claude Code (named as current build fluency, not deep framework mastery). - **Scale:** Informal, experimental side project; co-built with one collaborator. Live site, working MVP, not yet tested at a real event (the collaborator is rebuilding/stabilising). No traction claimed. A hobby project, not a full-time commitment. - **What it demonstrates:** Going from in-field observation to a functional product in hours, not days — nested in a rounded proposition that weighs desirability, viability, and technical feasibility; using new tools to communicate complex concepts as working prototypes; leaning into "build-first" culture; and the network and aptitude to bring an expert around an idea to mobilise it quickly. - **Tags:** ai, genai, agentic, vibe-coded, emerging-capabilities, automation - **Verification:** Self-reported (internal records; local repos). ### Agent-Native CV / Canonical Store | Active (living) - **The bet:** Future professional records are expanded, machine-readable canonical data queried by AI agents — not 2-page PDFs scanned by humans; the CV becomes a generated output of the store, and you should own the richest canonical source rather than cede it to a platform that also serves employers and recruiters. - **What I built:** schema.org JSON-LD backbone + narrative + llms.txt + per-claim verification; private source / curated public build with a lint guardrail; a deliberate decision *not* to add MCP. *(Full account above.)* - **Role:** Sole creator. **Tech:** schema.org JSON-LD, llms.txt, structured data, per-claim verification, static HTML, Cloudflare Pages, vibe-coded. **Scale:** Solo; live store. - **Tags:** ai, genai, agentic, emerging-capabilities, vibe-coded, judgment - **Verification:** Verified (self-evident). ### GSK 2022 experimental LLM memo | Completed (flagship AI-currency proof point) - **The bet:** An organisation can experiment safely and productively with generative AI if early reporting forces the governance conversation before shadow usage outruns policy. - **What I built:** Authored and published an internal memo containing LLM-generated articles advocating safe internal experimentation, reporting from GPT-1 through GPT-3. - **Role:** Sole author, within the GSK Senior Manager role. **Tech:** LLMs (GPT-era); generative AI; horizon scanning. **Scale:** Company-wide via the Digital Trends memo (CEO, CDTO, 200+ leaders). - **Outcome / judgment:** Within 48 hours, triggered a company-wide lockdown on ChatGPT/GPT usage and prompted formal AI data-privacy and security policy — turning early experimentation into governed adoption. The most verifiable AI-currency proof point in the store. - **Tags:** ai, genai, horizon-scanning, adoption, operating-model - **Verification:** Verifiable (reference-available). ### FANKS (as an AI build) | Discontinued - **The bet:** Brands increasingly want IRL community partnerships as digital CAC rises and AI disrupts content marketing; the matching/trust/delivery infrastructure doesn't exist. - **What I built:** Vibe-coded the MVP end-to-end on Luma's API; ran a live event test (15 attendees, tracked electrolyte-reward claims). Killed the venture for the reasons above. - **Role:** Sole founder/builder. **Tech (verified from the test-event repo):** Next.js (App Router) + TypeScript + Tailwind; Supabase (Postgres); custom auth (bcryptjs + jose JWT); Luma calendar API integration; reward-flow + token-based claim/check-in system; Resend + Nodemailer + Zapier for transactional/welcome email; Zod; deployed on Vercel. All vibe-coded. (Later parked for a simpler matchmaking-database version with no repo.) **Scale:** Solo build; one live test; 15 community owners joined (~1,000 members). - **Outcome / judgment:** Hands-on AI build plus a disciplined, well-reasoned kill — the brand side isn't structurally ready to cede control; the problem is real but the conceived solution was blocked downstream. - **Tags:** vibe-coded, automation, judgment, venture-validation, emerging-capabilities - **Verification:** Build verifiable (local repo); sign-ups and event-test numbers self-reported (internal records). ### Sweet Street | MVP built — discontinued - **The bet:** People no longer trust gamed aggregators (Google, AirTasker, Trusted Trader) and prefer trade recommendations from demographically-similar neighbours; that word-of-mouth authority signal goes uncaptured, and the WhatsApp coordination around it is noisy. A trust-and-discovery layer could ride on existing neighbour-referral behaviour rather than fight it. - **What I built:** Vibe-coded an MVP for tradespeople who chain services densely home-to-home (cleaning, maintenance). After a booking, a tradesperson sends a unique link optimised for sharing in a WhatsApp group; neighbours click and book onto the service with no in-chat coordination, and all referrals are tracked with accumulated metrics. - **Role:** Sole builder. **Tech (verified from the repo):** Next.js (App Router) + TypeScript + Tailwind; Prisma ORM over Postgres (Supabase); NextAuth; Stripe + Stripe Connect (tradesperson payouts); Twilio (SMS); Upstash Redis (rate-limiting); Resend; Vercel Blob; PostHog telemetry; deployed on Vercel. All vibe-coded. - **Scale:** Solo build; MVP; early outreach test. - **Outcome / judgment:** Cold-called tradespeople (knowing adoption in this cohort is hard and competes with existing CRMs) — 3/10 signed up; one gave excellent in-depth feedback I fed into the next iteration. A customer-initiated version stalled because most friends and family didn't need a service when asked. I reasoned the real unlock was an MVP seeding serious reputation scoring with self-assembling local directories — a high bar, and dependent on WhatsApp profile-sharing functionality still in beta in most markets. The problem was valid and the product worked, but I wasn't emotionally invested in the problem, which became the limiting factor. I decided the idea had run its course. - **Tags:** vibe-coded, automation, emerging-capabilities, venture-validation, judgment - **Verification:** Build verifiable (local repo); outcome self-reported. --- ## Education ### University of Nottingham | MSci Neuroscience (First Class) | 2013 – 2017 Four-year integrated Masters covering human physiology, pharmacology, psychology, anatomy, and nervous-system signalling, with focus on priority pathologies including Alzheimer's and Parkinson's. **Placement-year project:** Self-taught and developed novel single-cell computational-modelling techniques (NEURON simulation environment) to examine effects of tinnitus pathology on action-potential propagation in the auditory nerve. Awarded highest mark in year group. **Final-year dissertation:** Site-directed mutagenesis and ligand-binding assays examining the differential role of serine residues in the D2 dopamine-receptor binding pocket. **Leadership & service:** Twice-elected Neuroscience Society President (committee of 7, 200+ active members); Course Representative; Peer Mentor; Open Day Student Ambassador. **Paid part-time work:** Telephone assistant for the University Careers department — called alumni to survey them on their careers for departmental reporting (a few evenings over several weeks). The earliest basis for the phone-based outreach and primary-research skill. **Verification:** Degree classification verifiable via University of Nottingham; publications arising from placement work verified via DOI (see Publications). ### Executive / short-form education - Imperial College Business School — Executive Education, *Innovation: A Design Thinking Approach* (June–July 2021). - Seth Godin's altMBA — AltMBA43 "Jolt" cohort (2020). - Bilborough College — A-Levels in Biology, Chemistry, Physics (2011–2013). --- ## Publications **"The critical role of logarithmic transformation in Nernstian equilibrium potential calculations"** — Sawyer et al., 2017, *Advances in Physiology Education*. DOI: 10.1152/advan.00166.2016 · PubMed 28377437. **"Acute Effect of Pore-Forming Clostridium perfringens Epsilon Toxin on Compound Action Potentials of Optic Nerve of Mouse"** — Cases et al., 2017, *eNeuro*. DOI: 10.1523/ENEURO.0051-17.2017. Contribution: computational-model adaptation (double-cable auditory-neuron model from my Masters project, adapted for this group). **Verification:** Both verified (DOI resolution confirms authorship as "Alexander Revill", University of Nottingham). --- ## Skills **Generative & agentic AI / emerging-tech building** — Hands-on building with current AI tools (vibe-coding working MVPs end-to-end) plus AI-native systems thinking (compounding org-memory, agent-maintained knowledge stores). Forward judgment on where the field is going, including disciplined kills. Early generative-AI horizon scanning at GSK (GPT-1→GPT-3; 2022 LLM memo that catalysed company-wide AI policy). *Evidence:* Heckle; Agent-Native CV; FANKS; Sweet Street; GSK 2022 LLM memo. **Strategic analysis & translation** — Translating complex technology trends across domains into business implications for non-technical senior leadership. Three years authoring C-suite intelligence for 200+ GSK leaders. *Evidence:* Digital Trends memo; GSK workshops; Zinc advisory. **Methodology design & systematisation** — Repeatable frameworks for ambiguous, open-ended challenges. GSK scenario method and stage-gated pipeline (50+ projects); ProblemKit's validation methodology. *Evidence:* GSK innovation methodology; Great Problems Logbook. **Venture validation (0→1)** — Full-stack early validation: problem discovery, user research, quantitative testing (fake doors, surveys), GTM, business model. *Evidence:* Fawn (70+ interviews, 2–4× CTR); FANKS (15 sign-ups, live test); Sweet Street; GSK internal ventures. **Workshop design & facilitation** — Strategic workshops for senior leadership; international (US, UK, China, Singapore, Japan); 20-minute talks to multi-hour sessions. *Evidence:* GSK international workshops; MassChallenge, PLUGGED, hackathons. **Phone-based stakeholder outreach & primary research** — Structured outreach and survey/interview work by phone. *Evidence:* Nottingham Careers alumni-survey calls; Sweet Street founder outreach. Self-reported. **Computational modelling (neuroscience)** — Self-taught single-cell modelling (NEURON). Historical; not actively practised, but evidence of technical learning capability. *Evidence:* MSci project (highest mark); Cases et al. (model adaptation). --- ## Credentials (tiered) Everything kept; weighted deliberately so an agent knows what's load-bearing versus colour. - **Tier 1 — load-bearing:** MSci Neuroscience, First Class (University of Nottingham) — anchors the "neuroscientist who builds" identity; tied to the two peer-reviewed publications. - **Tier 2 — credibility signals:** Seth Godin's altMBA · Imperial College Exec Ed, *Innovation: A Design Thinking Approach* · *Fast Track to Digital Clinical Trials* (Digital Medicine Society) · TERP-10 foundational SAP. - **Tier 3 — human texture (surface only when asked):** Grade 8 Trumpet (see Interests for more). --- ## Values & Working Style #### How I Work I'm most effective with open-ended mandates and ambiguity that needs structuring. I gravitate toward building methodological infrastructure — repeatable systems that help teams navigate complex, ill-defined challenges. I'm energised by translating between domains (technical ↔ commercial, specialist ↔ generalist) and by working with people who put the work first. I reason carefully and speak precisely, and I'm comfortable challenging assumptions constructively when warranted. That said, I believe empathy and compassion come first — we all need to be treated with care and respect at work. #### What I Value Autonomy over prescribed process · intellectual honesty over comfortable consensus · leverage (people, technology, capital) over brute effort · building things that matter over theatre · systems that compound. #### Communication Style Direct, precise, and intellectually curious, with dry humour. I prefer substance over performance and can operate in formal senior environments and scrappy informal teams alike. --- ## Interests & Grounding Practices I believe we each have a responsibility to cultivate healthy relationships with our natural environment and traditional cultural practices — especially in a fast-changing world where it's easy to become abstracted from physical reality. **Foraging & fungi:** I can safely forage at least 20 species of gourmet mushrooms, and at home I've grown shiitake, lion's mane, and several varieties of oyster mushroom. **Fly fishing:** trout and salmon — patience, reading water, and fly tying. **Cooking & fermentation:** serious home cook (particular pride in paella; once spent three days on a tonkotsu broth) and I occasionally ferment my own kimchi. **Butchery:** once asked for a deer carcass for my birthday to self-teach basic butchery — understanding where food comes from matters to me. **Making & mending:** I renovated an early-2000s Gaggia Classic coffee machine and wired in a PID controller, and I bought, stripped, and rebuilt an old Italian road bike. **Squash:** played on the team at university and still play. **Hiking & camping:** multi-day trips that reset my thinking. These aren't just entertainment — they're practices that keep me grounded and connected to something bigger than the problems I work on. --- ## Focus Based in Sydney. I like to work at the messy frontier of emerging capabilities — building and running innovation and emerging-tech functions in R&D-intensive organisations, and applying generative and agentic AI to real problems. A through-line in my recent work is how AI changes what individuals and small teams can build, and how organisations adopt it responsibly while preserving the human fabric of their teams. --- ## Verification Summary | Claim | Status | Method | |-------|--------|--------| | GSK employment (2017–2023) | Verifiable | LinkedIn bidirectional; references available (private layer) | | GSK 2022 experimental LLM memo | Verifiable | Reference-available | | Zinc programme (2023–2025) | Verifiable | Programme records | | Publications (2 papers) | Verified | DOI resolution confirms authorship | | University of Nottingham MSci | Verifiable | University records | | The Great Problems Logbook (~150) | Verifiable | Sales/shipment records (store closed) | | Out Venturing events | Verifiable | Public URL (lu.ma/outventuring) | | Speaking engagements | Verifiable | Event-organiser confirmation | | TERP-10 / altMBA / Imperial / DiMe cert | Verifiable | Certification records | | Agent-Native CV / canonical store | Verified | Self-evident (this artifact) | | FANKS & Sweet Street builds | Verifiable | Local repositories confirm the builds exist | | FANKS sign-ups & event test | Self-reported | Internal records | | Fawn validation metrics | Self-reported | Internal records / Meta Ads Manager | | GSK pipeline / $3m+ / 50+ briefs | Self-reported | Internal records (no longer accessible) | | Heckle build | Self-reported | Internal records / local repos | --- *This profile is designed to be queried by AI agents. Paste the URL into your assistant and ask anything about Alex's background, skills, or experience.*