./projects

The brief was a prospect list. A long list of generic AI companies would have been easy to produce and useless to work, because every name on it needs a specific buyer and a reason that company would care.

Freelance GTM contractor. Spec build.

First, who actually counts as a prospect. Technical teams from pre-seed to Series A, roughly 5 to 50 people, whose own knowledge is scattered across Slack, GitHub, docs, and old incidents, and who have recently done something that suggests they are about to feel it: raised, hired, or shipped.

Then thirty accounts instead of three hundred, split into four groups. Each group is a different guess about who the product is really for, which makes the outreach a test rather than a queue.

Then the Airtable underneath it. Opportunities for where a deal stands and what happens next, accounts for whether a company fits, leads for the individual buyer and how they were reached, owners for who is responsible. The rules say what has to be known about a company before anyone contacts it, and what gets written down after every conversation.

Delivered as a spec build. Because each of the four groups tests a different guess, the outreach comes back with something useful about the market whether or not it comes back with meetings.

Women founders in India get about ₹4 of every ₹100 raised. Nobody had tested whether that gap is investors funding women less for the same business, or women founding in sectors and cities where rounds are smaller. Same gap, opposite policy fix.

Solo. Built the dataset, ran the analysis, wrote the paper.

Hand-collected 249 Indian startups from PitchBook, CB Insights, and LinkedIn, all with follow-on funding between 2023 and 2026. Tracked founder gender and CEO gender separately. Ran four specifications, adding sector and state fixed effects.

The raw gap is real: $16.3M average per round for male-founded, $7.4M for female-founded. Add sector and state fixed effects and the gender coefficients lose significance across every specification. Within already-funded firms, most of the gap runs through where women found rather than who funds them. The sample has seven female-founded firms and only includes companies that raised, which constrains everything above.

Government data on Delhi street vendors only counts licensed ones. Unlicensed vendors are the majority, so the official picture describes a minority of the sector.

Alone in the field. Designed the study, ran every interview, wrote the paper.

Surveyed 40 vendors across North, West, South, and Central Delhi over eight weeks, in Hindi, on foot. Chaat, momos, meal food, burgers.

10 of 40 held a mandatory license. 30 of 40 paid monthly bribes to police or MCD officials. 77.5% were migrants. The vendors with licenses were the ones with post-secondary education or 30 years of family experience, which means the licensing system selects for exactly the people who need it least.

The findings are in the process of being adopted by India's national street-vendor association.