Infrastructure & Inference Engineer

Systems that survive the spike.

I build production AWS and Kubernetes platforms, make GPU inference materially cheaper, and turn difficult operational paths into systems teams can trust.

10M+ TPM prepared for peak-traffic load testing
30% of production traffic served by self-hosted models
68% GPU infrastructure cost reduction
$230k annual cloud savings delivered for a banking client

01 / Selected systems

Measured outcomes, not infrastructure theatre.

System 01 · Scale

Peak-traffic platform readiness

Owned end-to-end load testing for revenue-critical traffic events, turning capacity limits into an actionable scaling plan across the production platform.

1.6M concurrent users 6M DAU and INR 4Cr revenue supported in 36 hours.

System 02 · Economics

Owned inference, lower unit cost

Deployed quantized diffusion models and Gemma via vLLM on dedicated GPU infrastructure, reducing dependency on third-party inference APIs.

INR 3.5–4 → INR 0.8 per inference, with self-hosted models serving 30% of production traffic.

System 03 · Real-time

Voice AI over private EKS

Built WebRTC media infrastructure with coturn behind an NLB, carrying SRTP/UDP audio into private-subnet EKS workloads for real-time AI calls.

Sub-second latency across the Android-to-AI voice path.

02 / Experience

From cloud migrations to production AI.

Five-plus years working across AWS, Azure, and GCP—with recent depth in platform scale, inference infrastructure, databases, and reliability.

Dec 2024 — Present

Senior DevOps Engineer

Primetrace

Own high-scale AWS infrastructure and production Kubernetes systems for consumer products; build self-hosted inference, real-time media, event-driven autoscaling, database reliability, WAF, and delivery automation.

Aug 2022 — Nov 2024

DevOps Engineer

Ollion

Delivered cloud cost optimization, GKE-to-EKS and cross-cloud data migrations, and automated security hardening across more than 80 AWS accounts.

Nov 2020 — Aug 2022

Senior Systems Engineer

Infosys

Migrated Redis and Go workloads to Azure-managed services and AKS with Terraform for major global financial institutions.

03 / Working depth

Built across the whole production path.

Cloud & platform

AWS, EKS, EC2, RDS, S3, Route53, CloudWatch, WAF, Azure AKS, GCP GKE and BigQuery.

Orchestration & IaC

Kubernetes, Terraform, ArgoCD, Argo Workflows, Docker, Helm, KEDA, GitHub Actions.

Inference & real-time

A100, L40S, H100, vLLM, Gemma, diffusion models, LoRA, WebRTC, coturn, TURN/STUN and SRTP.

Reliability & data

MySQL/InnoDB internals, Redis, load testing with Locust, observability, backups, access governance, Python and Bash.

1st place — AWS GameDay Led the top-scoring team in the Acquisition Meltdown / AWS DevOps Agent competition, finishing with 99,270 points among competing startups.

04 / Contact

Building infrastructure that earns its complexity.