Sarvam
|AI Engineering Intern
Highlights
Implemented two new APIs—Language Identification and Transliteration—and enhanced the Translate API with configurable modes and postprocessing options; currently maintain the text-models repository.
Collaborated with an open-source organization to integrate scalable, cost-efficient data generation pipelines for model training and generated data at scale for training new models.
Developed and maintained internal tools for model uptime monitoring and regex-based text replacement.
Contributed to the evaluation suite by designing prompts to assess production data across key parameters and expanded the benchmark dataset with failure cases for improved model benchmarking.