AI-Assisted Software Development at Scale

Personally built the infrastructure and scaled AI coding tools to 800 engineers in 6 months

Overview

I believed in the importance of AI-assisted development before anyone else at Appian was noticing it. I personally built the infrastructure, services, and rollout strategy to bring AI-assisted development tools to Appian's entire engineering organization. Starting from a prototype in Feb 2025, the program scaled to full adoption across 800 engineers within 6 months, achieving 5-40% productivity gains.

Results

  • 800 engineers using AI-assisted development tools daily
  • 5-40% productivity gains measured across code generation, test writing, and documentation
  • Prototype to full adoption in 6 months
  • Engineering Department Impact Award for cross-organizational AI adoption
  • Program became a model for how Appian rolls out new developer tooling

My Contribution

I implemented most of the infrastructure myself because I believed in this before the rest of the organization recognized its potential:

  • AWS Infrastructure (Terraform): Designed and deployed the production infrastructure for Amazon Q Developer using Terraform on AWS. Built separate dev and prod environments with proper state management (S3 + DynamoDB). Deployed through MARS CI/CD pipeline with strict production controls.
  • CI/CD Pipeline (MARS): Configured the MARS deployment pipeline for automated, safe production deployments. Developers apply changes locally for dev; production changes go through automated MARS pipeline with full audit trail.
  • Cost Management: Built user deactivation system to manage $19/user/month license costs. Integrated with AWS IAM Identity Center and Okta to programmatically identify inactive users and revoke access. Created cost dashboards for executive visibility.
  • Adoption Tracking: Built activity tracking pipeline within AWS to enrich raw telemetry logs with human-readable user information (name, email, team) from IAM Identity Center. Created adoption dashboards showing usage patterns across teams.
  • Group Configuration: Designed the group-based access model integrating Amazon Q with Okta SSO groups, enabling self-service provisioning and team-level controls.
  • Company-Wide Enablement: Delivered 3 Appian Talks driving adoption:
    • "Releasing Amazon Q" (Jul 2025) — Initial rollout announcement and onboarding
    • "Make Boring Less Boring with Q" (Aug 2025) — Practical use cases and productivity tips
    • "Context is King: Build Your Own Custom Assistant for Your Team" (Sep 2025) — Advanced usage, custom assistants, team-specific configurations
  • "Context is King" Workshop: Designed and delivered hands-on workshop that gave every engineering team their own custom AI assistant with team-specific context (codebase, architecture docs, testing patterns). Each team walked away with a highly performant AI agent that could generate accurate code, design architecture, and write tests specific to their domain. Partnered with AI Copilot group and Enterprise Automation team to scale across 3 groups — breaking down silos and creating a cultural shift. This directly accelerated Agent Studio and Composer delivery as most code and features were then written using these custom AI agents.

Impact Award

Engineering Department Head presenting the Impact Award for driving cross-organizational AI adoption.

© 2025 Sahil Gupta. All Rights Reserved.