gaurav sharma

• *Operate hybrid multi-cloud infrastructure for cross-cloud products and services — AWS and GCP alongside on-prem, including Kubernetes clusters that span all three environments.
• *Own the cross-cloud network fabric: site-to-site VPN tunnel architecture, routing, and connectivity between AWS, GCP, and on-prem.
• *Deploy and run self-hosted platform services end to end (Airflow, databases) — provisioning, upgrades, and reliability — where self-hosting beats managed offerings on cost or control.
• *Drive FinOps across both clouds: cost optimization folded into architecture decisions up front, not retrofitted after the bill arrives.
• *Own reliability, monitoring, and observability across cloud and on-prem systems; land defaults that are secure, compliant, and cost-effective at the same time.
• *Partner with ML teams on MLOps infrastructure: Dask clusters for distributed compute and Airflow-orchestrated model-training pipelines.
• *Build platform automation — from infrastructure workflows to onboarding scripts that turn manual setup into repeatable, auditable runs.
• *Embed AI-led development practices and workflows into infrastructure engineering, working across cross-region teams.

• *Scaled infrastructure from a single monolith to a network of 56 microservices as one of the earliest infra hires, owning platform growth from day one.
• *Drove FinOps practices that cut infrastructure costs 60% — roughly $100K/year in savings.
• *Migrated infrastructure to Terraform with 96% coverage.
• *Implemented DevSecOps controls: SSO, least-privilege cloud access, and periodic security & vulnerability audits.
• *Hardened Docker images for security and size; established VCS, branch-protection, review, and SSH-access processes to ISO compliance.
• *Introduced version-controlled infrastructure diagrams and partnered with developers to optimize code and queries for lower resource consumption.

• *Built a Kubernetes orchestration platform at Rakuten with a drag-and-drop UI for composing and orchestrating Kubernetes YAML.
• *Collaborated with a 21-developer team at RelaySolutions on a US-healthcare appointment & ride-booking product.
• *Worked extensively with Docker manifests and image layering.

• *Architected backends and scalable infrastructure for ~30 mobile apps across fitness, education, e-commerce, and ed-tech.
• *Led a team of 5 developers and 2 interns, mentoring them into full-stack engineers.
• *Owned end-to-end infrastructure for multiple clients and helped establish early-stage development processes, project management, and team coordination.

• *Stepped in for senior developers to close long-pending, high-priority modules — including a stalled payments module — on schedule.
• *Contributed to MultusMedical (Arizona-based), rendering DICOM medical images in the browser.

Model-agnostic coding agent with smart model-tier routing, a unified API-key layer, and prompt caching. Open-core: public cli-core via git subtree with GitHub Actions sync and ESLint import-boundary enforcement. Built under dugleelabs (dugleelabs.io).

Open-source reference implementation of adaptive-bitrate video streaming in Node.js: transcoding to Apple HLS (m3u8) for bandwidth-aware video-on-demand over HTTP. 80+ stars on GitHub, with a companion engineering write-up.