Worked on improving the stability and correctness of the autoscaling loop in the hashicorp/nomad-autoscaler repository, focusing on preventing negative scaling scenarios. Addressed a bug in the threshold scaling strategy by ensuring that when the instance count was zero, the newCount value would be clamped at zero, eliminating the risk of unintended negative scaling. Added regression tests to verify that newCount remains non-negative across edge cases, tightening the guardrails around scaling decisions. Utilized Go for development and testing, with changes aimed at reducing operational risk, preventing cost spikes, and increasing confidence in production autoscaling behavior for Nomad environments.