This dataset contains real-world COVID-19 human-to-human transmission network data. The data mimics how COVID-19 infections may have spread from one individual to another in Bucharest (Romania) during August 1st and October 31st, 2020. The information refers to COVID-19 patients (referees) and their contacts (referrals), i.e., the people they interacted with before being tested COVID-19 positive. The dataset is structured as an edge-list file (referee - referral ties). For each referee (referral), we provide the following attributes: sex (male/female), age, sector (public/private), a job in the medical sector (yes/no), ISCO-08 one-digit code, ISCO-08 two-digit code, ISCO-08 three-digit code, employability (active/non-active), age class (minor, adult, pensioner), confirmation month (when a patient was tested positive for COVID-19 infection), confirmation day (when a patient was tested positive for COVID-19 infection). The data were analyzed using relational hyperevent modeling (https://github.com/juergenlerner/eventnet). This dataset allows replication of the analysis reported in the manuscript entitled: Occupations and their impact on the spreading of COVID-19 in urban communities (Hâncean M-G, Lerner J, Perc M, Oană I, Bunaciu D-A, Stoica AA & Ghiță M-C).