# Mohammed Saaduddin Ansari

> Principal SRE · Cloud & AI Platform Architect, based in Hyderabad, India. I build and run AI agents, MCP servers and AIOps pipelines in production. 11+ years running production platforms on GCP, AWS and Azure at LivePerson, SADA, EPAM, SAP Labs, Deloitte and Infosys. These days I build AI agents that triage incidents, cloud foundations that survive bad days, and platforms engineers actually like using.

Goes by Saad. This site is his career written as an engineering operating manual (Doc. SRE-001, Rev. 14, revised 13.11.2026). Every fact below is generated from the same data file as the page.

## Availability

- Status: Accepting new workloads. Standby · ready to deploy.
- Between deployments and available immediately. Open to Principal and Staff roles, and to fractional, consulting, teaching and co-founder work in SRE, cloud architecture, platform engineering and AI platforms. Hyderabad, remote-first. Open to relocation (US, Canada, UK, Singapore, Gulf) with visa sponsorship.
- Primary goal: Principal / Staff engineer (Full-time · immediate start).
- Also available as: Fractional cloud, AI or SRE architect; Fractional CTO; Consulting & freelance projects; Workshops & courses; Career advice & mentoring; Technical co-founder.
- Working hours: Overlap with US, EU or APAC hours as needed. Ack within 24 hours.
- Contact: LinkedIn https://www.linkedin.com/in/md-saaduddin-ansari (the email address is shown on the site to human visitors only).

## Results

- MTTR: −50% (LivePerson · AIOps pipeline; INC-007)
- On-call noise: −45% (LivePerson · 500+ alerts/day correlated; INC-007)
- Uptime: 99.99% (Fortune 500 · multi-region, multi-cloud; INC-006)
- Recovery time: 4h to <20m (Banking system · disaster recovery; INC-006)
- Apps migrated: 300+ (SADA · to GCP and AWS; INC-006)
- Cloud cost: up to −40% (SADA · FinOps for clients; INC-006)
- Deploy cycle: 2 weeks to 4 hours (SAP Labs · CI/CD; INC-004)
- Years in production: 11+ (6 companies, zero regrets)

## Systems run in production

- AI agents & agentic workflows (since 2023; LivePerson): Agents that triage alerts, investigate incidents and run runbooks. MCP servers give them tool access, a human approves every action, and evals check their output before release. Proof: 500+ alerts a day grouped into incidents; −45% on-call noise. Stack: MCP, LangGraph, LangChain, CrewAI, AutoGen, OpenAI Agents SDK, Human-in-the-loop.
- LLM integration, RAG & evals (since 2023; LivePerson · SADA): LLMs wired into internal systems with tool use, structured outputs and routing between providers. RAG pipelines and multi-agent workflows are tested against golden datasets before they ship. Proof: 3 model providers, routed; 15+ client engagements using LLM-assisted IaC. Stack: Anthropic API, OpenAI API, Gemini API, pgvector, LangSmith, Prompt caching, Golden datasets.
- Cloud architecture & migrations (since 2015; SADA · SAP Labs · Infosys): Landing zones and foundations built from scratch, migrations planned end to end, and multi-region, multi-cloud designs on GCP and AWS that hold up on bad days. Proof: 300+ applications migrated; 99.99% uptime, Fortune 500 client. Stack: GCP, AWS, Azure, Terraform, Landing zones, GKE, EKS.
- Disaster recovery (since 2022; SADA): Recovery designed and rehearsed before it is needed. For a critical banking system that meant cutting recovery from hours to minutes and losing almost no data. Proof: 4h → <20m recovery time (RTO); 1h → <5m data loss window (RPO). Stack: Multi-region, RTO / RPO design, Kasten K10, Terraform, Runbooks.
- Observability & incident response (since 2020; LivePerson · SADA · SAP Labs): SLOs and error budgets agreed with the business, telemetry on OpenTelemetry, and ML anomaly detection that pages a person only when it matters. Proof: −50% MTTR at LivePerson; 15 → <3m mean time to detect. Stack: OpenTelemetry, Prometheus, Grafana, Datadog, SLOs, AIOps.
- Kubernetes & delivery (since 2017; SAP Labs · EPAM · Deloitte): Clusters across GKE, EKS and AKS, and pipelines that make releases boring: GitOps, drift that reverts itself, and rollback without a war room. Proof: 2w → 4h deployment cycle, zero downtime; 50+ deploys a day with auto-rollback. Stack: Kubernetes, Helm, Argo CD, GitHub Actions, Jenkins, Spinnaker, Istio.
- Internal developer platform (since 2021; LivePerson · EPAM): Self-service infrastructure through templates and golden paths, so service teams ship without filing a ticket and waiting a week. Proof: 50+ engineers on the platform; days → <30m environment provisioning. Stack: Backstage, Crossplane, Terraform modules, Golden paths, Policy as code.
- FinOps (since 2022; SADA · LivePerson): Tagging policy, commitment planning and right-sizing, with AI-assisted recommendations, so the cloud bill stops being a surprise. Proof: up to −40% client infrastructure cost; −35% approx., after a 100+ app move to GCP. Stack: Tagging policy, Commitments, Right-sizing, Kubecost, CAST AI.
- Security, compliance & AI guardrails (since 2022; SADA · LivePerson): Landing zones that stay inside SOC 2 and HIPAA controls, a framework that took a healthcare agency to Authority to Operate, and guardrails for AI agents. Proof: ATO reached for a healthcare agency; SOC 2 & HIPAA-aligned landing zone. Stack: IAM, OPA / Gatekeeper, Vault, Prompt-injection guards, PII redaction.

## Experience (11+ years)

### Principal Site Reliability Engineer, LivePerson (Jun 2025 – Nov 2026)

- 500+ alerts a day. Most of them noise, all of them paging a human.
- No correlation, no context. Engineers were pattern-matching dashboards at 3 a.m.
- Built an AIOps pipeline on OpenTelemetry with ML anomaly detection, plus AI agents for triage, investigation and runbooks. Every agent action waits for a human to approve it.
- Rolled out a Backstage + Crossplane developer platform for 50+ engineers and LLM-assisted Terraform review, so infrastructure changes shipped noticeably faster.
- On-call noise down 45%. MTTR down 50%. Mentored a team of 8 along the way.

### Cloud Architect, SADA (An Insight Company) (May 2022 – May 2025)

- 300+ applications to move to GCP and AWS, and a critical banking system that needed 4 hours to recover.
- Built GCP foundations from scratch, planned the migrations, designed multi-region architecture and disaster recovery. Brought LLM-assisted Terraform and PR review to client work from 2023.
- Banking RTO from 4h to under 20m, RPO from 1h to under 5m. 99.99% uptime, multi-region and multi-cloud, for a Fortune 500 client. Infrastructure costs down by up to 40% across 15+ engagements.

### Lead DevOps Engineer, EPAM Systems (Nov 2021 – May 2022)

- GitOps with GitHub Actions and ArgoCD, so drift gets detected and reverted on its own. A reusable Terraform module library.
- Environment provisioning from days to under 30 minutes. Led a team of 6 through 4 engagements with zero critical incidents.

### Sr DevOps Engineer, SAP Labs (Jan 2020 – Nov 2021)

- Owned the GKE clusters as Product Owner across 100+ environments. CI/CD with Jenkins, Spinnaker and Ansible. Standard Prometheus and Grafana monitoring.
- Deployment cycle from 2 weeks to under 4 hours, zero downtime. MTTR from 4 hours to 30 minutes.

### DevOps Engineer, Deloitte USI (Oct 2017 – Jan 2020)

- Terraform and Ansible for 10+ enterprise clients: 2 days → 30 minutes. Jenkins pipelines running 50+ deploys a day with automatic rollback.

### DevOps Engineer, Infosys (Jun 2015 – Oct 2017)

- Helped move 20+ client applications from data centres to AWS. Automated away 30+ hours of manual work a week.

## Resume

- [Datasheet (PDF): Principal SRE](https://mohammed-saaduddin-ansari.pages.dev/resume/Mohammed_Saaduddin_Ansari_Principal_SRE.pdf): the primary resume, two pages.
- [Resume, AI platform emphasis](https://mohammed-saaduddin-ansari.pages.dev/resume/Mohammed_Saaduddin_Ansari_AI_Platform.pdf): Same history, different ordering.
- [Resume, Cloud architecture emphasis](https://mohammed-saaduddin-ansari.pages.dev/resume/Mohammed_Saaduddin_Ansari_Cloud_Architecture.pdf): Same history, different ordering.
- [Resume, Platform engineering emphasis](https://mohammed-saaduddin-ansari.pages.dev/resume/Mohammed_Saaduddin_Ansari_Platform_Engineering.pdf): Same history, different ordering.
- [JSON Resume](https://mohammed-saaduddin-ansari.pages.dev/resume.json): the same history as structured data.

## Other editions

- [The manual](https://mohammed-saaduddin-ansari.pages.dev/): the full page.
- [metrics.txt](https://mohammed-saaduddin-ansari.pages.dev/metrics.txt): results in Prometheus exposition format.
- [healthz.txt](https://mohammed-saaduddin-ansari.pages.dev/healthz.txt): status in ten lines.
- [LinkedIn](https://www.linkedin.com/in/md-saaduddin-ansari)
- [GitHub](https://github.com/mohammed-saaduddin-ansari)