Skip to content

Sai Dheeraj Gantala

Backend Engineer · Agent Platforms · Bengaluru

Building agent platforms with MCP·LangGraph·Temporal·Google ADK·Kubernetes·Backstage

Read the case study →Notes on agent reliabilityStack →Career arc
§02 / 09

Career arc

  1. Oct 2025 — Present

    System Engineer — Enterprise Agent Platform

    EPAM Systems

    Bengaluru, India

    • Building an enterprise AI agent platform — MCP integrations, agentic workflows, operator review loops bridging AI and engineering automation.
    • Data integration pipelines, multi-tenant RBAC, and role-based dashboards for developer self-service + operator approval flows.

  2. Jul 2023 — Sep 2025

    Software Engineer — Cloud, Data & DevOps

    Premium Parking (Xebia)

    Hyderabad, India

    • 30% revenue lift via end-to-end PostgreSQL→MSSQL data pipeline (AWS DMS + custom triggers powering near-real-time BI).
    • 80% backup-time out via Bash parallel-processing automations; 40% system-performance lift on restore workflows.

  3. Jul 2022 — Jun 2023

    Associate Software Engineer — Backend & Microservices

    Premium Parking (Xebia)

    Hyderabad, India

    • Decoupled a monolith into microservices using Hasura GraphQL + React.js.
    • Migrated Sidekiq background jobs to AWS Lambda in Python — 30% memory out, higher concurrency.

  4. Mar 2022 — Jun 2022

    Engineer Intern

    Xebia

    Hyderabad, India

    • Tripled deployment frequency via GitHub Actions CI/CD; Dockerized apps + deployment validation, 30% deploy-time reduction.
    • Hands-on Go, Docker, Java, Rails, Python, AWS, GCP across backend and cloud projects.
§03 / 09

Currently building

active

Enterprise Agent Platform @ EPAM

MCP-integrated agent workflows, multi-tenant RBAC, Temporal orchestration, LangGraph + Google ADK agents. Operator review and approval loops built in.

PythonFastAPILangGraphTemporalGoogle ADKBackstage
active

Hiiired + Noxstack (personal)

AI-powered recruiting (RAG + CrewAI) and community local-service discovery (FastAPI + LangChain). Solo weekend builds.

FastAPINext.jsRAGOpenAILangChain
§04 / 09

Certifications

§05 / 09

Multi-cloud DevOps

I've shipped production workloads across AWS, Azure, and GCP. Pick a provider to see what I built there.

AWS

  • PostgreSQL → Azure MSSQL pipeline via AWS DMS + custom triggers — +30% revenue lift on BI dashboards.
  • 35% AWS cost out via rightsizing + Lambda lifecycle automation (CloudWatch Events); 65% provisioning time out via Terraform + AWS CDK.
  • AWS WAF + reCAPTCHA cut fraudulent activity 70%; ELK + Elastic APM cut MTTD / MTTR 30%.
  • AWS Cognito (OIDC) integrated with Ruby on Rails + Azure AD for SSO.
§06 / 09

Ventures

Side projects and products I've shipped or am currently running outside of my day job.

Hiiired

active

Solo builder · 2024 — Present

AI-powered recruitment: resume tailoring, job matching, authentication, workflows.

FastAPINext.jsRAGOpenAICrewAI
Visit →

Noxstack

active

Solo builder · 2024 — Present

Community-driven local service discovery — find and book neighborhood pros.

FastAPIPostgreSQLAWSTerraformLangChain
Visit →

WeDAA

active

Co-founder · 2023 — Present

Cloud-native app prototyping + backend code generation. Scaffold to deploy in minutes.

Spring BootTerraformKubernetesAWS
Visit →
§07 / 09

Principles

Tap a principle to read the reasoning behind it.

AI agents should be observable, debuggable, and interruptible. Black boxes are for demos, not production.
Most of what I know about agent reliability came from incidents, not documentation. I optimize for fast feedback loops.
A 3x cost reduction is usually a 3x latency reduction with better architecture. FinOps is engineering.
Postgres, queues, and well-understood patterns handle 90% of agent platforms. The novel 10% is where AI lives.
§08 / 09

Recognition

§09 / 09

Get in touch

Send a note about what you're building. I read everything and reply within a couple of days.

Open mail →
or use copy buttons above
Sai Dheeraj Gantala — AI Agent Platform Builder