Krishna Mankali

Review AWS cloud infrastructure and automation tasks to assess technical correctness, security, operational realism, and AI-agent performance.

• Analyze task requirements, setup and solution scripts, Python graders, teardown procedures, and execution transcripts.
• Independently evaluate agent runs against the intended outcome, identifying reasoning gaps, incomplete repairs, and discrepancies between grader results and actual behavior.
• Assess least-privilege access, encryption, event-driven pipelines, monitoring, configuration drift, and safe resource cleanup across AWS services.
• Review Terraform and AWS CDK configurations for dependency handling, resource preservation, and consistency with deployed infrastructure.
• Provide evidence-backed feedback, precise citations, and actionable recommendations to improve task clarity, reference solutions, and validation coverage.
• Maintain a structured tracker to document findings, compare recurring failure patterns, and identify duplicate task designs.

• Infrastructure Automation:
  Designed and deployed AWS infrastructure with CDK for Terraform (CDKTF), managing VPC, subnets, security groups, autoscaling groups, IAM, ALB, and CloudWatch monitoring.
  Automated builds and deployments with cdktf synth/get/deploy pipelines.
  Wrote unit and integration tests (Jest) to validate stack resources before deployment.
• LLM Training (RLHF):
  Contributed to training large language models by curating, annotating, and ranking model responses for Reinforcement Learning with Human Feedback (RLHF) tasks.
  Evaluated model behavior against quality, accuracy, and alignment guidelines to improve model reliability.
  Provided structured feedback to enhance prompt completion, reasoning, and user intent alignment.

Tech Stack: CDKTF, Terraform, AWS (EC2, VPC, S3, IAM, CloudWatch, ALB, AutoScaling), TypeScript, Jest, LLMs, RLHF, Prompt Engineering.

Role: AWS DevOps Engineer
Ekaggata Technology Consultant Pvt Ltd – Bangalore (March 2022 – Feb 2023)

• Designed and deployed AWS infrastructure using Terraform and CloudFormation.
• Developed CI/CD pipelines using Jenkins, Maven, GitLab CI/CD, and AWS CodePipeline.
• Containerized applications with Docker and deployed using Kubernetes (EKS) and Helm.
• Integrated SonarQube for static code analysis.
• Managed artifacts using JFrog Artifactory and Nexus.
• Monitored systems using Prometheus, Grafana, and CloudWatch.
• Automated infrastructure with Ansible.
• Collaborated with dev and ops teams to optimize performance.