Worked on the kvcache-ai/sglang repository to deliver a Tokenizer Special Tokens Configuration feature, enabling precise and flexible preprocessing for the model gateway. Addressed tokenizer encoding issues in the Go bindings by implementing a targeted fix, which improved reliability and reduced runtime errors during tokenization. Leveraged Golang and Rust to enhance the tokenization pipeline, focusing on configuration-driven token handling and cross-language integration. The work resulted in more accurate preprocessing and cleaner integration with the model gateway, demonstrating a strong grasp of backend development, tokenization, and Go bindings while prioritizing maintainable, configurable solutions to address edge cases in encoding.