Dmitrii K.

AI/ML Infrastructure Architect | Production AI on AWS | 1M+ Earned

1M+ earned on Upwork building production AI/ML infrastructure for VC-backed startups and enterprise teams. Previously: Platform Engineer at Unstructured (open-source document ETL for LLMs and RAG pipelines, YC-backed) and Stack AI (enterprise AI workflow platform). 100% Job Success Score | 12,343 hours delivered | 102 projects completed AWS Certified Solutions Architect | San Francisco, CA Recent engagements: - Senior Platform Engineer (Python, AWS, Kubernetes) - 24K contract, 284 hours. Built GitOps pipeline reducing deployment cycles from 2 hours to 15 minutes. Maintained 99.9% uptime across production EKS clusters. - AWS backend for a Forbes-recognized VC-backed SaaS. Architected infrastructure handling 10K+ concurrent API requests, reduced AWS costs by 35% through right-sizing and reserved instance optimization. - Azure Virtual Desktop Architect and Technical Lead - designed multi-region AVD deployment for 500+ internal users. - Production data migration and infrastructure modernization - 15K contract, 218 hours. Migrated 50TB+ across cloud environments with zero downtime, automated data validation reducing manual QA by 80%. What I build: - Production AI/ML pipelines on AWS (EKS, Lambda, SageMaker) - Multi-agent systems and MCP server deployments on Kubernetes - Self-hosted LLM infrastructure (Llama, open-weight models) - ML model serving, monitoring, and inference optimization - Terraform IaC, GitOps pipelines, CI/CD for AI workloads - Cloud cost optimization for GPU-heavy and inference environments How I work: I partner with CTOs and engineering leads who need AI infrastructure that scales without breaking. No boilerplate. Every architecture decision is documented, versioned, and handed over so your team can maintain it independently. Engagement formats: - AI Infrastructure Audit (fixed scope, 5-day turnaround) - Full platform build (4-12 week engagements) - Ongoing AI/ML platform engineering (monthly retainer, async-first) Available now. Click Invite to Job to discuss your architecture.