About the Role
This role involves building a proof-of-concept prototype from end to end for a next-generation automated code review tool powered by large language models.
Responsibilities
- Design and implement an LLM-integrated code review system.
- Analyze pull requests, detect bugs and security vulnerabilities, and provide contextual feedback within CI/CD pipelines.
- Own the entire proof-of-concept (PoC) process.
- Manage system architecture, LLM integration, static analysis pipeline, pull request webhook integration, review dashboard, and Docker-based deployment.
- Integrate LLM APIs (e.g., OpenAI GPT-4, Anthropic Claude) into production systems.
- Develop and maintain CI/CD pipelines using GitHub Actions and GitLab CI.
- Incorporate static analysis tools (e.g., Semgrep, CodeQL) into the review workflow.
- Work independently from initial architecture through to deployment.
Requirements
- 5+ years of full-stack development experience with Python or Node.js and React or Vue.
- Hands-on experience integrating LLM APIs (e.g., OpenAI GPT-4, Anthropic Claude) into production systems.
- Strong background in CI/CD pipeline development.
- Experience with GitHub Actions, GitLab CI, and code review tooling.
- Experience with static analysis tools (e.g., Semgrep, CodeQL).
- Ability to work independently and take ownership of the entire PoC lifecycle.
- Familiarity with Retrieval-Augmented Generation (RAG) architectures and vector databases (plus).
Skills
- Full-stack development
- Python
- Node.js
- React
- Vue
- LLM integration
- OpenAI GPT-4
- Anthropic Claude
- CI/CD pipeline development
- GitHub Actions
- GitLab CI
- Code review tooling
- Static analysis tools
- Semgrep
- CodeQL
- System architecture
- Docker-based deployment
- Retrieval-Augmented Generation (RAG) architectures
- Vector databases
Location
- San Francisco
- Remote-friendly
Work Type
- Contract
Experience Level
- 5+ years of full-stack development experience
About the Company
- NexaFlow is developing a next-generation automated code review tool powered by large language models (LLMs).
