Varun Kumar B

▸ Developed and maintained Terraform modules and AWS CloudFormation templates to provision and manage scalable AI/ML platform infrastructure across multi-tenant environments, ensuring reproducible and auditable cloud deployments. ▸ Architected AWS Lambda and Step Functions workflows to orchestrate end-to-end ML inference pipelines, enabling serverless, event-driven AI workloads with fault tolerance and automatic retry logic. ▸ Designed IAM roles, policies, and permission boundaries to enforce least-privilege access across AI platform components, ensuring compliance with enterprise security standards and regulatory requirements. ▸ Deployed DynamoDB tables and S3 data lakes to support AI platform state management, model artifact storage, and real-time feature retrieval for ML inference at scale. ▸ Built scalable cloud-native ML serving infrastructure on AWS EKS with auto-scaling and high availability, integrating with RDS, S3, and Lambda for asynchronous inference and model lifecycle management. ▸ Supported full-stack AI platform management applications in Python (FastAPI), enabling monitoring dashboards and ML-driven anomaly detection, reducing manual operational intervention by 30%. ▸ Communicated technical dependencies, deployment impacts, and change request evaluations to scrum team, vendors, and business stakeholders, ensuring transparent delivery and cross-functional alignment. ▸ Established CI/CD pipelines (GitHub Actions, Jenkins) for automated infrastructure provisioning, model builds, and deployment with rollback capabilities, ensuring production reliability of AI services. ▸ Integrated Prometheus, Grafana, and OpenTelemetry for end-to-end observability of AI platform workloads, defining SLOs and proactive alerts to identify and resolve performance bottlenecks.

▸ Provisioned and managed AWS cloud infrastructure using Terraform and CloudFormation to support HIPAA-compliant AI/ML platform for clinical analytics, ensuring scalability, reliability, and security across multi-organization environments. ▸ Designed and deployed AWS Lambda and Step Functions orchestration workflows for clinical ML pipelines, including patient scheduling optimization and provider workload forecasting models. ▸ Configured IAM policies and RBAC controls for secure, compliant access to clinical AI platform resources across AWS services (EKS, S3, RDS, Lambda), meeting strict healthcare regulatory requirements. ▸ Leveraged DynamoDB and S3 for scalable storage of ML model artifacts, inference results, and healthcare feature stores supporting real-time clinical decision support systems. ▸ Built Python (FastAPI) backend microservices to expose ML model inference as secure, high-performance REST APIs, supporting integration with real-time clinical dashboards and EHR workflows. ▸ Deployed containerized ML microservices on AWS EKS with Helm and Kustomize, leveraging autoscaling, workload security policies, and integration with S3 and Lambda for zero-downtime model updates. ▸ Established CI/CD automation (GitHub Actions, Jenkins, ArgoCD) for ML model versioning, testing, and zero-downtime deployments, reducing release cycle times by 40%. ▸ Evaluated change request impacts on shared cloud infrastructure components and coordinated with platform teams and vendors to resolve technical conflicts and deployment dependencies.

▸ Designed and deployed RAG-based financial knowledge retrieval systems and AI agents using LangChain, LlamaIndex, and OpenAI APIs, delivering contextual AI-powered insights across structured financial datasets. ▸ Built and managed knowledge bases for intelligent document search and financial data retrieval, enabling AI agents to surface regulatory and transaction insights with high accuracy across enterprise data stores. ▸ Authored Terraform infrastructure code and AWS CloudFormation stacks to provision and manage Lambda functions, Step Functions workflows, DynamoDB tables, S3 buckets, and IAM roles for AI platform components at a Tier-1 US bank. ▸ Orchestrated complex AI/ML workflows using AWS Step Functions to coordinate multi-step financial data processing, model inference, and regulatory reporting pipelines across distributed cloud resources. ▸ Enforced IAM least-privilege policies and VPC security configurations across all AI platform components, ensuring SOC 2 compliance and adherence to financial industry security standards. ▸ Deployed event-driven, fault-tolerant ML workloads on AWS (EKS, Lambda, Glue, MSK/Kafka, S3), enabling real-time financial event streaming and ML-powered alerting with 40% query latency improvement. ▸ Built asynchronous background ML job processing using Celery and Redis, improving batch inference throughput by 35% and reducing blocking dependencies across downstream analytical systems. ▸ Collaborated with scrum teams and stakeholders to communicate technical implications of AI platform changes, evaluate impact of new requirements, and persuade cross-functional partners on infrastructure design decisions.

• Designed PostgreSQL schemas with tenant-based data partitioning for marketplace analytics and multi-region reporting, improving query performance by 30% and supporting scalable multi-tenant reporting requirements. • Migrated monolithic services to containerized microservices, accelerating deployment speed by 60% and enabling independent service scaling underpinned by structured SDLC change management and rollback documentation. • Built ETL ingestion workflows processing structured partner data (JSON/XML) into PostgreSQL with validation and reconciliation pipelines, reducing data quality issues by 55% across partner integrations. • Led API routing and error-handling refactor initiative, reducing post-deployment production incidents by 70% and improving MTTR across critical booking services through structured root cause analysis. • Authored technical documentation and architectural diagrams for API workflows and service dependencies, reducing onboarding time and eliminating knowledge silos across multidisciplinary engineering teams.

Relevant Courses: Algorithms, Advanced Database Systems, Data Mining, Topics in Computer Science: Data Science

Relevant Coursework: Programming for Problem Solving, Operating Systems, Computer Architecture, OOPs with Java, Computer Networks, and Cloud Computing.