The short answer
LDelight builds and operates DevOps platforms for New York companies on AWS and Azure. We set up CI/CD, infrastructure as code and monitoring so releases become routine, and we design the pipeline so it doubles as the change-management evidence SOC 2 auditors and NYDFS examiners ask for. Our working day is your night, which suits overnight releases.
Key takeaways
-
Change management evidence for SOC 2 comes straight out of the pipeline
-
Privileged access is MFA-gated and time-limited, in line with NYDFS Part 500
-
Your asset inventory is generated from infrastructure-as-code state, not a spreadsheet
-
Overnight Eastern release windows fall inside our normal working hours
Release often, prove every change
New York companies selling to banks, insurers and large enterprises meet the same question in every security questionnaire: how do you control change? A SOC 2 Type II report answers it with evidence — that changes were authorised, tested and approved, consistently, across the audit period. Teams that build that evidence by hand spend weeks before each audit reconstructing tickets and screenshots.
We set the pipeline up so the evidence exists by construction. Branch protection enforces review; the pipeline records which commit and artefact reached production, who approved it, and which tests ran; infrastructure changes are planned, reviewed and applied by the pipeline rather than a person in a console. Your auditor samples deployments and finds the same shape every time. That consistency is what turns a stressful audit into a routine one.
Privileged access built for Part 500
The amended NYDFS cybersecurity regulation now requires multi-factor authentication for all remote and privileged access and a maintained asset inventory. Both shape how a DevOps platform is built. Humans should rarely need privileged access at all, because deployments run under a pipeline identity scoped to what it deploys. When someone does — during an incident — access is requested, approved, MFA-gated and expires on its own, through AWS IAM Identity Center or Entra Privileged Identity Management. Break-glass accounts exist, are monitored, and are tested.
The inventory follows from infrastructure as code. If every resource is declared in Terraform, Bicep or CDK and drift is detected, the state files are an accurate, current list of what you run — far more reliable than a spreadsheet updated once a quarter.
Asynchronous by design
We do not overlap New York business hours, and we do not pretend to. Work is handed over in writing at the end of our day, and a fixed 08:00 Eastern call covers anything that needs a conversation. The gap is an advantage for release engineering: overnight deployment windows, database migrations and rollback rehearsals happen while our team is fully at work and your traffic is lowest.
What you get, and what you keep
A typical engagement begins with a two-week assessment of one or two services and ends with pipelines, infrastructure code, dashboards and runbooks in your repositories and accounts. Nothing depends on tooling we own. Many clients then keep us on a monthly operations retainer; others take it in-house. Either path is planned from the start.
Page reviewed Nov 14, 2026. Region and regulatory facts for the US checked Nov 14, 2026. Rules change — confirm current obligations with your adviser.