Research Intern — Compilers Domain (IIT Hyderabad)
Built an ML-driven bytecode vector embedding pipeline (bytecode2vec) to support program analysis and compiler optimization research. Applied compiler IR analysis techniques to identify optimization opportunities from real-world bytecode corpora. Benchmarked ML models for code understanding using program analysis outputs from LLVM IR-based pipelines. • Developed static analysis routines that extract bytecode patterns for downstream ML • Used performance profiling and benchmarking to evaluate model and pipeline effectiveness • Worked within LLVM IR and program analysis frameworks to generate/validate analysis artifacts • Focused on ML model evaluation applied to compiler optimization tasks