How are VC Decisions Made?
Speaker event · Oxford Venture Capital Network · 2026
What actually happens behind closed doors when a VC decides to back a startup or pass on it.
Alongside my research I work across early-stage venture capital: running the venture community at Oxford, investing into deep-tech and AI-native companies, and building quantitative tools for sourcing and diligence.
I am President of the Oxford Venture Capital Network (OxVCN), Oxford's venture capital society. The Network connects students with founders and investors through speaker events, educational VC workshops and its flagship Venture Capital Programme, which trains aspiring analysts.
I lead the events programme, using my own and the society's network to bring investors and operators to Oxford, and running workshops and hackathons in partnership with other university societies.
I am an investor at Proximo Ventures, an early-stage fund backing deep-tech founders out of the top UK universities (Oxford, Cambridge and Imperial) across climate tech, life sciences and Industry 4.0. The model puts investors on the ground at the universities they cover: sourcing companies, running due diligence and working directly with founders.
I am also an angel investor in Assuric, one of our portfolio companies.
I scout for Ellipsis, a Zurich-based fund writing first cheques into AI-native companies at pre-seed and seed. Ellipsis is led by Yariv Adan, co-founder of Google Assistant and Google Lens across seventeen years at Google, alongside founding partners including Matthias Dantone, and runs its sourcing, diligence and operations on an autonomous agentic system.
The mandate is companies where AI is the product rather than a feature on top of one, with defensibility coming from proprietary data and deep domain expertise.
I was a research intern at Vela Partners, a San Francisco fund that describes itself as the first product-led, AI-powered venture firm, and which publishes much of its AI-for-VC research with the University of Oxford.
My work there applied quant-finance backtesting discipline (point-in-time data, leakage controls, survivorship-free universes, pre-registration) to venture signals: predicting which research ideas compound rather than fade, and which founder profiles produce outliers. Vela now runs an extended version of the first of these in house as a monthly research-trends product.
Speaker events, workshops and hackathons run with the Oxford Venture Capital Network. The society's full listing, including upcoming events, is at oxfordvcnetwork.com/events.
Speaker event · Oxford Venture Capital Network · 2026
What actually happens behind closed doors when a VC decides to back a startup or pass on it.
Speaker event · Oxford Venture Capital Network · 2026
Turning the pitch around: leading VC firms presented their investment thesis, sector focus and approach to partnering with early-stage founders, to a curated room of Oxford entrepreneurs.
Networking · Oxford Tech Week · 2026
Connecting potential co-founders at Oxford Tech Week, on the basis that choosing one is among the most consequential decisions a founder makes.
Hackathon · Oxford · 2026
An all-female gamified hackathon over two days in Oxford, co-hosted with the Oxford Founders Society, the OX1 Incubator and others.
Networking · The Lighthouse Pub, Oxford · 2026
An informal evening combining pitch conversations with Oxford Seed Fund office hours, deliberately held in a pub rather than a lecture theatre.
Fireside chat · Oxford Venture Capital Network, with the Oxford China Group · 2026
A fireside chat with Oxford alumnus Adam Knight, co-founder of YASO, following the company’s $11M Series A.
Speaker panel · Oxford Venture Capital Network · 2026
A panel on the funding disparities female founders face, followed by speed networking between founders and investors.
Speaker event · Keble College, Oxford · 2026
Is the current wave of AI investment genuine innovation, or the next tech bubble? A discussion at Keble College.
Speaker panel · Oxford Venture Capital Network · 2026
An in-person panel with founders and venture operators on investor relations and what it takes to secure VC backing in health tech.
Fireside chat · Oxford Venture Capital Network · 2026
An in-person fireside chat and open Q&A with Alexander Schmitt, Partner at Lightspeed Venture Partners, a firm with over $40B under management and 101 unicorns.
Quantitative research on venture signals, built during and after my research internship at Vela Partners. Every project below applies the same discipline: point-in-time data, a leakage test that fails the build, survivorship-free universes, and hypotheses fixed before scoring.
Research project · Vela Partners · 2026


Most research ideas decay, a few compound. This reads a new idea’s first months on arXiv and predicts whether it will still be growing two years later. It achieves +0.23 AUPRC over momentum, with six of six pre-registered cross-field predictions correct.
GitHub Repository: https://github.com/judepops/halflife-research
Research project · Vela Partners · 2026


Modelling founder credentials as a typed attribute graph and clustering the GNN embeddings into interpretable archetypes, giving a 29.7 pp success-rate spread across a 9,000-founder benchmark, 6.9x that of tabular clustering.
GitHub Repository: https://github.com/Vela-Research/graph-founder
Research project · Vela Partners · 2026
A market-level, longitudinal, leakage-free feature corpus for predicting whether a market, independent of any team, produces an outlier company. The market-level sibling of Vela’s founder-level research.
GitHub Repository: https://github.com/Vela-Research/market-brain