Newvick Lee activity https://gitlab.com/nlee8 2026-03-18T22:54:22Z tag:gitlab.com,2026-03-18:5219738482 Newvick Lee commented on epic #21290 at GitLab.org 2026-03-18T22:54:22Z nlee8 Newvick Lee [email protected]

Yes that's right, behind an environment variable setting. But the environment variable isn't recommended for production setting, so that's a problem.

tag:gitlab.com,2026-03-18:5218784474 Newvick Lee commented on epic #21290 at GitLab.org 2026-03-18T17:23:45Z nlee8 Newvick Lee [email protected]

We recently had an escalation with a customer, and one of the proposals is

Expanded DAP/AI Gateway test coverage for WebSocket/streaming reliability and self-hosted scenarios

Would that be in the scope of the end2end test or a health check?

tag:gitlab.com,2026-03-17:5214944936 Newvick Lee opened issue #593926: Create dynamic GPU requirements calculator at GitLab.org / GitLab 2026-03-17T22:02:53Z nlee8 Newvick Lee [email protected] tag:gitlab.com,2026-03-17:5214906209 Newvick Lee opened issue #593925: [Config - Level 2] Extend Duo Health Check with vLLM checks at GitLab.org / GitLab 2026-03-17T21:53:16Z nlee8 Newvick Lee [email protected] tag:gitlab.com,2026-03-13:5203163471 Newvick Lee commented on merge request !225703 at GitLab.org / GitLab 2026-03-13T23:13:09Z nlee8 Newvick Lee [email protected]

Just reading the MR to learn from it since I was originally assigned the issue...

The health check improvement is a very nice touch

tag:gitlab.com,2026-03-12:5198739031 Newvick Lee commented on merge request !140 at GitLab.org / AI Powered / Custom Models / Custom Models 2026-03-12T20:41:17Z nlee8 Newvick Lee [email protected]

Good point, updated !140 (3b4cb0c0)

tag:gitlab.com,2026-03-12:5198737044 Newvick Lee pushed to project branch nlee8-main-patch-9998 at GitLab.org / AI Powered / Custom Models / Custom Models 2026-03-12T20:40:42Z nlee8 Newvick Lee [email protected]

Newvick Lee (3b4cb0c0) at 12 Mar 20:40

Update file getting_started.md

tag:gitlab.com,2026-03-12:5198387200 Newvick Lee approved merge request !226809: docs: add AWS Bedrock BYOM deployment guide to solutions at GitLab.org / GitLab 2026-03-12T18:46:01Z nlee8 Newvick Lee [email protected]

Summary

  • Add a step-by-step guide for deploying GitLab Duo Self-Hosted with AWS Bedrock as a BYOM provider
  • Covers EC2 provisioning, AI Gateway setup, TLS/gRPC configuration, model assignment, and DAP flow verification on a single-instance Docker deployment
  • Cross-linked from the self-hosted models admin docs page (doc/administration/gitlab_duo_self_hosted/_index.md)
  • Complements the existing Ollama integration guide

What's included

  • Full end-to-end walkthrough: blank EC2 instance to working DAP flow on Bedrock
  • Copy-pastable commands for every step
  • Architecture diagram (Mermaid)
  • NGINX reverse proxy config for HTTPS + gRPC TLS
  • Bedrock-specific gotchas: inference profile IDs (us. prefix), Anthropic use case form, Converse API bug
  • Troubleshooting section covering 6 common failure modes
  • Verified against GitLab source code (env vars, admin UI paths, Rake tasks, port defaults) and real deployment experience

Files changed

File Change
doc/solutions/integrations/aws_bedrock_byom.md New guide (~800 lines)
doc/solutions/integrations/_index.md Added link under "Self-Hosted Model"
doc/administration/gitlab_duo_self_hosted/_index.md Added cross-reference in "Related topics"

Verification

  • markdownlint-cli2: 0 errors
  • Verified against doc/install/install_ai_gateway.md, doc/administration/gitlab_duo_self_hosted/configure_duo_features.md, doc/administration/gitlab_duo_self_hosted/supported_llm_serving_platforms.md, and ee/lib/tasks/gitlab/duo.rake
  • Verified against real deployment logs from multiple Bedrock BYOM deployments

Test plan

  • Review renders correctly on the docs site (check Mermaid diagram, tables, note blocks)
  • Verify all relative links resolve correctly
  • Technical review by someone who has deployed BYOM with Bedrock