code-translation

Simple repo showing code-to-code and code-to-text capabilities using LLMs on Databricks.

(WDLIC) - a system that takes natural language descriptions and converts them into both pseudo-code and executable code across multiple domains

LangGraph agent that translates Python functions to JavaScript with sandboxed verification and a bounded debug-retry loop

MoLE (Mix-of-Language-Experts) model architecture for multilingual programming

Qwen3.5-27B fine-tuned with a three-stage Rust, debugging, and C2Rust curriculum; 87.3% execution-verified success on 200 C programs.

☰ FutЯøn ☰ | Reverse-engineer, customize, and add features to any installed app. Patches compiled code, injects custom UI, translates code between 30+ languages. LLM-agnostic — works with Ollama, Claude Code, OpenClaw, or any AI assistant. Zero cloud cost.

Use the power of LLMs to automagically ID and fix code errors, refactor and improve code quality, and translate code between languages along with auto-sourced substitute imports

Argos Hiragana Coder is a Python utility that translates Python-like pseudo-code written in Japanese (Hiragana) into executable Python code using English syntax using the Argos Translate Library

Translates multi-file PHP projects to Python using LLM agents — trusts the difference, not the model.

AI-powered code translator: Code <> English <> Code. Built on Groq + DeepSeek with triple LLM failover, real-time SSE streaming, Razorpay billing, and zero-budget Render deployment.

Benchmark and evaluation harness for LLM-based parallel code translation across CUDA, OpenMP, OpenCL, and OpenMP target offload (NeurIPS 2026)

RAG-based C/C++ to Python translator using local LLMs, AST parsing, and semantic code search

AI-powered code translation across 6 languages using LangGraph multi-agent orchestration, RAG with Qdrant, and Claude Sonnet

Fortran-to-Python weather forecasting application translation with OOP design — M.S. Computer Science coursework

RIIR-style harness + skill for reliable C→Rust and Python→Rust ports on mid-tier LLMs. Staged milestones, frozen-test anti-cheat, polish. Measured on OpenCode free MiMo + Nemotron (subset fixtures — not full FastAPI/inih/Bun).

A benchmarking framework for evaluating LLM-based COBOL-to-Java converters. Tests performance of converters with equivalence tests and code quality metrics.

Eight open, tested skills for AI coding agents: verified code translation (C/Rust/asm) with deadlock analysis and safety-critical evidence, skill routing, cross-agent pipelines, quality loops, citation-verified research, and 2D-look Blender animation via MCP

LLM-augmented legacy Perl/SQL → Python translation with behavioral equivalence proof: dual execution, cell-level diff, CI gate at 99%, generated regression suites. 85.7% auto-translation, 99.63% equivalence.