This study investigates the impact of social identity on judicial processes and outcomes at the Patna High Court over a decade (2009 to 2019). We employ machine learning algorithms to infer caste status from surnames (names) in court records. We note that a majority of court participants have ‘caste-neutral’ names. Though we find no evidence of name-based ‘matching’ between litigants and judges, caste-neutral petitioners are 3.3% more likely to choose a caste-neutral advocate. Matching, whether intentional or coincidental, yields notable consequences. Litigants with caste-neutral names who matched with similar judges face a higher likelihood of case dismissal and lower success rates in overturning appeals. However, advocates with caste-neutral names experience less disadvantage in these scenarios. Notably, the adoption of caste-neutral names, while offering some protection, does not fully mitigate the vulnerabilities faced by citizens within India's judicial institutions.