A comprehensive programming language compiler engineered from first principles. The system implements complete lexical analysis with token recognition, recursive descent parsing with robust error recovery, AST-based constant folding optimization, and LLVM IR code generation for efficient cross-platform compilation.
View ProjectA Python SDK for safely executing AI-generated code inside Firecracker microVMs. Spins up a fully isolated Linux environment in ~50ms via snapshot-based startup, supporting Python, C++, and Go. Features a pre-warmed VM pool, vsock communication, file system API, and detailed execution metrics — built specifically for LLM agent pipelines.
View ProjectWeb-based C++ memory leak detection platform powered by PLT hooking and E2B sandbox technology. Automatically detects memory allocation/deallocation mismatches including malloc/free, new/delete, and array leaks. Provides detailed stack traces, execution reports, and runs securely in isolated sandbox environments.
View ProjectA comprehensive system call tracing ecosystem with modified QEMU emulator and multi-level trace analyzer. Supports RISC-V and x86_64 architectures with real-time monitoring capabilities, deep syscall parsing, and thread-aware analysis for kernel debugging, security analysis, and performance profiling.
View ProjectAI-powered CPU profiling skill for automated performance analysis. Integrates Intel VTune Profiler and Intel Advisor CLI to identify hotspots, threading bottlenecks, vectorization opportunities, and memory access patterns. Provides automated diagnostic reports with optimization recommendations.
View Project