Skip to content
View sudeepr34's full-sized avatar

Block or report sudeepr34

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
sudeepr34/README.md

Sudeep Reddy

Site Reliability Engineer @ PhonePe · Bengaluru

Primary on-call for payment platforms serving 700M+ users. Previously ran 24×7 ops for Visa's 450 PB Hadoop estate (Altimetrik). Day-to-day: incident response, multi-region HA/DR, observability, and cutting toil.

  • PhonePe (Dec 2025 – present) — active-active MySQL Galera across Azure + on-prem, production Kubernetes, SLOs / error budgets, chaos testing
  • Altimetrik / Visa (2023 – 2025) — HDFS / Hive / Presto, Airflow (500+ daily pipelines), Terraform + Ansible, Kerberos / LDAP / IAM
  • Also — OCI 2025 DevOps Professional · MBA Data Science (Amity, 8.4/10)

Portfolio · sudeepreddy340@gmail.com


Stack

Cloud AWS (EC2, S3, Route 53, IAM, CloudWatch, EMR, VPC), Azure, OCI
Orchestration Kubernetes, Docker, Podman, Helm, Drove
IaC Terraform, Ansible, SaltStack
CI/CD GitLab CI/CD, Jenkins, Harbor
Observability Prometheus, Grafana, ELK, Splunk, OpenSearch, PagerDuty
Data MySQL, Percona, Galera, Aerospike, Hadoop, Hive, Presto, Spark
Scripting Python, Bash, Airflow, Control-M
Networking TCP/IP, DNS, BGP, VLAN/VXLAN, NGINX, NAT, firewalls

Projects

Prod work stays behind the firewall. These are the public versions.

Transformer-Based Anomaly Detection — TensorFlow/Keras transformer autoencoder over multi-metric telemetry. Trained on healthy traffic; reconstruction error is the score. Test: precision 0.95, recall 0.83, ROC AUC 0.97. Returns per-metric attribution so a detection says which metric broke, not just that something did.

Incident RCA Pipeline — Correlates Prometheus/ELK alerts into severity-ranked clusters, tags them as origin / correlated / separate, drafts runbook steps from the alert text. Optional LLM summary (Ollama or any OpenAI-compatible endpoint) with prompt redaction and a deterministic fallback if the LLM fails.

ansible-roboshop — 11-service stack end to end with Ansible (EC2, Route 53, MongoDB, Redis, MySQL, RabbitMQ, app tiers, systemd, NGINX).

shell-roboshop — Same stack in Bash — what a deploy looks like before you pull in Ansible.

terraform — Remote state, loops, conditionals, data sources, multi-env layouts.

Portfolio — Static site, no framework / build step. Strict CSP, security headers.

Currently

Running Gemma offline in a home lab with a log stream, as a local incident-response assistant that drafts RCA steps without data leaving the box.

Popular repositories Loading

  1. shell-roboshop shell-roboshop Public

    RoboShop stack deployed with Bash scripts.

    Shell

  2. ansible-roboshop ansible-roboshop Public

    11-service RoboShop stack provisioned with Ansible (EC2, Route 53, MongoDB, Redis, MySQL, RabbitMQ, app tiers).

    Jinja

  3. terraform terraform Public

    Terraform patterns: remote state, loops, conditionals, data sources, multi-env.

    HCL

  4. terraform-multiple-env terraform-multiple-env Public

    Terraform layout for promoting infra across env.

    HCL

  5. terraform-aws-instance terraform-aws-instance Public

    Minimal Terraform module for an EC2 instance.

    HCL

  6. LLM-Powered-Incident-RCA-Pipeline LLM-Powered-Incident-RCA-Pipeline Public

    Correlate Prometheus/ELK alerts into severity-ranked clusters and draft runbook steps. Optional LLM summary (Ollama or OpenAI-compatible) with offline fallback.

    Python