Resume
Highlights
Designed & built Unbxd’s tech stack from ground up
- Query (Search/Recommendations) Pipeline, serving 160+ million requests daily.
- Analytics Pipeline, handling 170+ million requests daily.
- Ingestion Pipeline, handling 1TB data daily.
- Inter-Region, inter-datacenter communication for pipelines.
Designed AI & ML Pipeline
- Datalake built on top of S3 & Parquet.
- Custom Orchestrator on top of Kubeflow + Kubernetes.
Contributor to multiple services currently in production
- Custom Edge Servers handling index-co-location, inter-region routing.
- Query analysis engine using custom grammar protocol.
- Low latency service for incoming request classification.
- Standardized libraries used at Unbxd (eg: https://github.com/unbxd/go-base).
- Project Starter codebases for Java & Go.
- Inter region event communication using NATS.io
Other contributions
- Gitops based CI/CD pipeline for service deployment using Flux CD
- Adoption of Protobuf as communication standard across the stack (In Progress).
- Serves as chief arbitrator on Architecture Review Board at Unbxd
Build & maintains self-hosted homelab, custom dofiles
- Homelab using ansible https://github.com/uknth/homelab
- Dotfiles https://github.com/uknth/dotfiles
Currently working on
- New Reporting Pipeline, using Apache Flink, Clickhouse/Big Query & Apache Superset.
- In-House A/B testing Framework based on ideas from Trustworthy Online Controlled Experiment