As a GenAI Solutions Engineer (MLOps-leaning) at eCloudvalley, you will turn AI prototypes into secure, observable, and cost-efficient production solutions on AWS. You'll collaborate closely with the Senior Cloud Architect, product / PS teams, and customers to build RAG / KB chat, voice assistants, and AI integrations—complete with evaluation loops, guardrails, and CI / CD.
Job Responsibilities :
- Design, implement, and operate serverless GenAI applications on AWS (Bedrock, Lambda, API Gateway, DynamoDB, S3, Step Functions, EventBridge, SQS).
- Productionize Bedrock Knowledge Bases / RAG flows and voice / chat assistants with clear SLIs / SLOs.
- Build evaluation frameworks (quality / latency / cost), and run scheduled regressions with dashboards.
- Implement guardrails & safety : PII redaction, content filtering, prompt hardening, audit logs, red-team tests.
- Establish observability by default : structured logging, metrics, alarms, and runbooks.
- Own cost management : infra budgeting, caching strategies, usage caps, FinOps tagging / alerts.
- Ship CI / CD pipelines (GitLab CI / GitHub Actions) and reusable Terraform modules for repeatable delivery.
- Collaborate on reviews (IAM / KMS, privacy, compliance) and incident readiness; support sales / PS with demos and workshops.
- Document architectures, playbooks, and "paved road" templates.
- Familiar in Data Pipeline / ETL – S3 + Glue + Athena + Step Function
Job Requirements :
Education / Experience : Bachelor's in CS / EE / SE (or equivalent). 3+ years on public cloud; 2+ years with AWS serverless or data / ML pipelines.Cloud & Security : AWS fundamentals—VPC / IAM / KMS, CloudWatchGenAI Platforms : Practical experience with AWS Bedrock (Models / KB / Guardrails) or equivalent LLM platforms.Application Eng. : Python (boto3) and / or ; API design, testing, troubleshooting.RAG & Data : Embeddings, chunking, retrieval tuning; OpenSearch or Aurora pgvector (or similar).MLOps for GenAI : Build eval harnesses (quality / latency / cost), A / B tests, monitor model / prompt changes.Integrations : Programmable Voice / SMS or Amazon Connect, webhooks, OAuth / OIDC