Open Source · Self Hostable · Docker Sandboxed

Code Execution
Infrastructure

For AI Applications

Run untrusted Python, C++, and Java in isolated Docker containers. Built for coding agents, interview platforms, and developer tools.

Docker · Go · Redis · MIT Licensed

The Stack

Built on battle-tested technologies

Go

HTTP API & worker pool

Docker

Execution sandbox

Redis

Job queue & persistence

MIT

License

Capabilities

Everything required to execute code safely at scale

01

Container Isolation

Every execution runs inside a disposable Docker container destroyed on exit.

02

Batch Evaluation

Compile once and run against multiple test cases concurrently.

03

Async Job Queue

Redis-backed execution workers for non-blocking workloads.

04

Security Constraints

CPU, memory, filesystem, network, and PID controls on every run.

05

API Key Management

Authentication, quotas, and rate limiting out of the box.

06

Self Hosting

Deploy anywhere Docker can run. MIT licensed.

Architecture

Simple architecture. Production-ready execution.

A linear pipeline from HTTP request to sandboxed container — no magic, no vendor lock-in.

01ClientYour application
02API GatewayReverse proxy / TLS
03Compiler ServiceGo HTTP API
04Redis QueueJob persistence
05Worker PoolConcurrent executors
06Docker SandboxIsolated containers
07Resultsstdout · stderr · exit code

Sandbox

Secure by default

Every constraint is applied on every run — there is no unrestricted execution path.

No Network Access--network none
Memory Limited--memory 256m
CPU Limited--cpus 0.5
Read-only Filesystem--read-only
PID Restricted--pids-limit 50
Non-root User--user 65534:65534
Ephemeral Containers--rm
Automatic Cleanupon exit

Interface

Simple API. Predictable responses.

REST endpoints with consistent JSON. Sync, async, and batch modes.

curl -s -X POST localhost:8080/v1/execute \
  -H "Content-Type: application/json" \
  -d '{
    "language": "python",
    "code": "print(input())",
    "stdin": "hello world"
  }' | jq

Use Cases

Built for teams shipping code execution

AI Coding Assistants

Let LLMs write and run code with strict sandbox boundaries.

Technical Interview Platforms

Evaluate candidate solutions against hidden test cases at scale.

Online Judges

Compile once, run many — batch endpoint built for competitive programming.

Learning Platforms

Give students a safe runtime without managing infrastructure.

Autonomous Coding Agents

Execute agent-generated code with predictable, auditable results.

Internal Developer Tools

Self-host on your VPC. Your code never leaves your network.

Self Hosting

Open source. Self-hostable. MIT licensed.

Go from clone to running infrastructure in under 60 seconds.

Terminal
$ git clone https://github.com/SaishKorgaonkar/compiler-service
$ cd compiler-service
$ make pull-images
$ make docker-up

Built By

Saish Korgaonkar

Final Year Computer Science student focused on backend systems, developer infrastructure, AI applications, and scalable software engineering.

Give your AI the ability to execute code safely

Self-hostable infrastructure for coding agents, interview platforms, and developer tools.