Mithil Acharya

Evaluate LLM outputs for accuracy, relevance, safety, and factual correctness using RLHF principles to support model alignment. Perform data preprocessing, annotation, and quality verification for text, image, and video to support AI/LLM training and validation. Tag, classify, curate, and verify multimedia datasets against detailed project guidelines, maintaining consistency across large volumes of unstructured data. Execute end-to-end annotation workflows on the Multimango platform and own data integrity for assigned batches prior to model ingestion. Conduct video annotation and bounding-box labeling for computer-vision/multimodal training datasets, ensuring guideline-aligned outputs and reliable labeling quality.

Designed and executed 150+ test cases across unit, integration, system, and field testing to support multiple major software releases in Agile sprints. Identified, triaged, and reported bugs/regressions across functional and regression cycles, driving issues through structured defect workflows to resolution. Uncovered usability issues during pre-launch testing and collaborated with developers and system teams to validate changes and confirm seamless integration before release. Led production verification testing with cross-functional engineers to reduce compliance issues and streamline release testing, applying manual black-box and exploratory testing as part of a broader regression strategy.