Project

AI Workflow Automation for Engineering

Built AI-assisted workflows for an engineering team, including automated PR reviews, Sentry issue triage with draft PR creation, AI-powered metrics insights, and LLM-based knowledge systems for internal documentation and operational context.

Overview

The engineering team had valuable information spread across pull requests, monitoring and metric tools, databases, logs, wikis, and internal documentation. This project focused on using AI tools and agents to connect those systems into practical workflows that reduced manual work, improved visibility, simplified business metrics access, and helped the team move faster without removing human review or ownership.

Technical details

A set of AI-assisted workflows and internal tools built for one engineering team, covering automated code review, Sentry-to-PR automation, an AI-powered Metrics Insights site, and an LLM-powered knowledge base wiki. Every workflow keeps a human in the loop before code or changes reach production.

  • AI Agents and LLM-based workflow orchestration
  • CodeRabbit for automated pull request review
  • Sentry integration for issue ingestion and triage
  • Automated draft PR creation for human review and approval
  • AI-powered Metrics Insights website over PostgreSQL and OpenSearch
  • LLM-powered knowledge base wiki over internal docs and runbooks
  • Retrieval over wikis, internal docs, and operational knowledge sources
  • Governance: scoped credentials, controlled data access, human-in-the-loop review

Challenges

  • Code reviews were becoming a bottleneck for the team
  • Bugs and production issues required repetitive investigation and manual fixes
  • Sentry issues needed triage before they became actionable engineering work
  • Business and non-business metrics were spread across PostgreSQL, OpenSearch, dashboards, and internal docs
  • Knowledge was fragmented across wikis, docs, and team-specific sources
  • AI had to be introduced with practical workflows, human review, clear ownership, and safe execution boundaries

What changed.

  • Automated PR reviews via CodeRabbit reduced reviewer overhead while keeping humans as final approvers
  • Sentry-to-PR agents turn production issues into draft pull requests for engineers to review, approve, and deploy
  • AI-powered Metrics Insights site unifies PostgreSQL, OpenSearch, wikis, and internal knowledge bases behind natural-language questions
  • LLM-powered knowledge base wiki improves access to runbooks, engineering decisions, and process documentation
  • Practical AI adoption model connected to real engineering workflows, with human review and maintainable engineering practices
  • Clear governance around what AI agents can read, suggest, or modify, with controlled access to internal systems

Want to talk about your product?

If you're setting up the foundations for your product and your team, let's have a 20-minute chat.

Book a 20-minute chat