yogesh@dreamstreet : ~/portfolio — zsh
[ 0.001s]BOOT initializing terminal session
[ 0.012s]CONN establishing connection to portfolio
[ 0.021s]AUTH yogesh@dreamstreet verified
yogesh@dreamstreet › ~/portfolio
whoami --verbose
roleAVP, Engineering · Platform, Cloud & DevOps
atDreamStreet AI (since Feb 2026)
basedMumbai, IN · IST (UTC+5:30)
tenure15+ years building infra that holds up at scale
domainsfintech (trading, payments) · gaming · production GenAI · developer platforms
cursor
instances
2,000+
aws fleet · daily peak
throughput
4,000+ tps
dreampay payments
users served
5M+
upstox trading platform
microservices
50+
k8s · zero-downtime deploys
yogesh@dreamstreet › ~/portfolio
cat ./currently.md
01 currently
AVP, Engineering · DreamStreet AI
Leading Cloud, Security, and DevOps for DreamStreet AI from a blank page. Designing the AWS account / IAM / networking baseline, CI/CD substrate, and platform security posture for the new entity. Architecting and building a multi-runtime Internal Developer Platform, applying the patterns proven with Odin at Dream11, with CLIs and self-service productivity tooling to replace a legacy EC2-only in-house deployment tool with a paved-road experience across EKS, ECS, and serverless. In parallel, solo author and operator of Niyam AI (production GenAI / RAG for stock-broker compliance, live 5+ months on AWS ECS Fargate and in daily use by a compliance and operations team) and building Code Wiki Battle Mode (multi-agent AI dev platform on Claude via AWS Bedrock).
yogesh@dreamstreet › ~/portfolio
ls -la ./case-studies/
02 case studies

CS_NOW DreamStreet AI — IDP & Cloud function from zero [ACTIVE]

problem
New legal entity for Dream11 group's next-generation AI work. No infrastructure, no platform team, no developer tooling. Existing in-house deployment tool is EC2-only and won't scale to multi-runtime workloads (EKS + ECS + serverless).
approach
Stand up Cloud, Security, and DevOps from a blank page: AWS account / IAM / networking baseline, CI/CD substrate, observability, platform security posture. Architect and build a multi-runtime Internal Developer Platform with self-service CLIs, paved-road deployment workflows, standardised secrets, observability, and SLO instrumentation across EKS, ECS, and serverless. Bake AI-assisted development (Claude Code, Copilot) into the build itself.
stack
AWSEKSECSTerraformGitOpsInternal Developer PlatformCLIsClaude Code
outcome
→ in active build (Feb 2026 → present) · platform-as-product operating model from day one

CS_01 DreamPay — zero to 4,000 TPS, PCI-DSS ready

problem
Greenfield payments platform. Required PCI-DSS readiness and high-throughput scale from day one. No prior infra, no team, no patterns.
approach
Architected complete AWS infrastructure, CI/CD pipelines, deployment workflows from the ground up. Single point of ownership for every infra decision.
stack
AWSEKSTerraformArgoCDPackerDatadogPCI-DSS
outcome
→ 4,000+ TPS sustained · PCI-DSS readiness achieved · production-stable since launch

CS_02 Dream11 — Odin, multicloud deployment platform [OPEN SOURCE]

problem
Deployments ran on a sprawling Jenkins job catalogue. Minor changes to the deploy process rippled across every job, the jobs couldn't configure VM or Kubernetes deployments, and there was no unified product vision for the lifecycle. During marquee tournaments, waiting on a new job to be built was a risk the business couldn't take.
approach
Led Odin, the platform that replaced it. Developers declare a Service Definition — components, versions, dependencies — and a provisioning file that picks the deployment flavour per component. A CLI does the rest: odin create env, odin deploy service, odin operate component. An extensible component registry let teams build and share their own components rather than queueing on the platform team, and cloud "flavours" made the same definition deployable to AWS or GCP.
stack
AWSGCPEKSEC2TerraformPackerArgoCDGitOpsMySQLRedisAerospikeSpark
outcome
→ 3,250+ on-demand environments at 6.28s average spin-up · 10,500+ dev and 4,030 production deployments · 22+ components · multicloud AWS + GCP · adopted by 520+ developers · open-sourced under Dream Horizon · read the write-up

CS_02b Dream11 — resiliency as a product

problem
Inconsistent SLAs across services on a 2,000+ instance fleet serving millions of concurrent users at sporting peaks. Drift between environments.
approach
Standardised immutable infrastructure with Packer + Terraform. Built the full observability stack. Deployed platform apps (monitoring, ingress, cert-manager, secrets) on EKS via GitOps with ArgoCD. CI/CD across 20–30 microservices.
stack
AWSEKSPackerTerraformArgoCDPrometheusGrafanaDatadogELKGitHub Actions
outcome
→ ~85% of deployments shipped without platform-team involvement · cycle time ~60% faster · incident MTTR ~40% lower · drift eliminated

CS_03 Kotak Neo — zero to trading-platform DevOps

problem
No DevOps practice. Manual deployments. On-prem infrastructure. Team had to launch a next-gen retail trading platform under SEBI/NSE compliance.
approach
Built the practice from zero. Designed Azure DevOps CI/CD across all internal projects. Led on-prem→AWS migration. Internal IaaS via Terraform. Ansible config across dev/staging/prod.
stack
Azure DevOpsAWSTerraformAnsibleSEBI/NSE
outcome
→ provisioning days → minutes · full CI/CD adoption · regulator-ready

CS_04 Upstox — scaling to 5M+ users

problem
Live retail trading platform scaling from early stage to 5M+ users. Needed cloud migration and zero-downtime deployments during market hours.
approach
Led on-prem→AWS migration of all critical trading systems. Migrated 50+ services to Kubernetes. Auto-scaling for market open/close spikes. Jenkins + GitLab CI for live deploys.
stack
AWSKubernetesJenkinsGitLab CITerraformAnsiblePythonLXDDocker
outcome
→ zero-downtime live trading deploys · 5M+ users served · cost cut via right-sizing/RIs
yogesh@dreamstreet › ~/portfolio
grep -i 'stack' ./skills.yaml
03 stack matrix
cloud
AWS (EKS, ECS Fargate, EC2, ALB/NLB, RDS, DynamoDB, Lambda, API Gateway, IAM, IAM Roles Anywhere, Bedrock) · Azure
iac & config
Terraform · Packer · Ansible · Helm
orchestration
Kubernetes · EKS · Docker · LXD
ci/cd & gitops
GitHub Actions · ArgoCD · Jenkins · GitLab CI · Azure DevOps · Internal Developer Platform
observability
Prometheus · Grafana · Datadog · ELK / OpenSearch
languages
Python (FastAPI) · TypeScript (Next.js, Fastify) · Bash · Go (familiar)
ai & agentic
Claude on AWS Bedrock · LlamaIndex · Qdrant · multi-agent orchestration · Claude Code (power user) · Copilot
security
PCI-DSS · SEBI / NSE · IAM Roles Anywhere · least-privilege design · break-glass production access
yogesh@dreamstreet › ~/portfolio
git log --oneline --reverse
04 experience timeline
2026 — NOW
AVP, Engineering @ DreamStreet AI
Cloud + Security + DevOps from zero · multi-runtime Internal Developer Platform · paved-road self-service tooling
2021 — 2026
Director, Engineering (Platform & Cloud) @ Dream11 & DreamPay
Led Odin, the multicloud deployment platform now open source under Dream Horizon (520+ developers) · reliability for 2,000+ AWS instances · DreamPay zero-to-PCI · IAM Roles Anywhere + break-glass · OKR Champion
2020 — 2021
Deputy Vice President @ Kotak Securities
Built DevOps from zero for Kotak Neo · 5–10 reports · on-prem→AWS
2019
Independent Consultant @ Upstox · LearnApp · Seal Infotech · Mangosent · MarketWolf
AWS architecture · CI/CD setup · cost optimization
2015 — 2020
Sr Solutions Architect @ Upstox (RKSV Securities)
Trading platform from early stage to 5M+ users · 50+ services to k8s
2013 — 2015
Technical Consultant @ OS3 Infotech
Open-source delivery for enterprise clients across India + ME
2011 — 2013
Linux Engineer @ OS3 Infotech
RHEL/SUSE production systems · core infra services
yogesh@dreamstreet › ~/portfolio
cat ./contact.txt
05 contact
email
bodke.yogesh@gmail.com
linkedin
linkedin.com/in/bodkeyogesh
phone
+91 99304 46046