Progressive Canary Lab
Zero-downtime deploys, health-gated rollouts, and instant kill switches — a hands-on system for shipping risky code safely, the way production teams do it.
A selection of things I've built - from open-source infrastructure to AI platforms and web apps.
Zero-downtime deploys, health-gated rollouts, and instant kill switches — a hands-on system for shipping risky code safely, the way production teams do it.
A production-inspired multi-agent platform that automates incident triage, root cause analysis, and remediation planning by orchestrating specialized AI agents across logs, metrics, deployments, and operational knowledge.
An end-to-end LLM system, engineered — data pipelines, RAG, fine-tuning, and a production inference API, wired together with ZenML, Qdrant, and AWS SageMaker.
An interactive distributed systems simulator that demonstrates how consistent hashing enables scalable data partitioning with minimal key redistribution. Visualizes virtual nodes, hash rings, and real-time cluster scaling.
An AI agent that watches live production metrics and decides — on its own — whether to promote, wait, or roll back a canary release. No hardcoded thresholds, just GPT-4o-mini reading error rates and latency in real time.
A Spring Cloud microservices platform for e-commerce — centralized config, service discovery, and an API gateway routing traffic to independently deployable user, order, and payment services.
A Quora-style Q&A platform broken into user, QnA, moderator, and admin services — with an API gateway routing requests and Kafka handling asynchronous processing under the hood.
A production-inspired agentic AI platform showcasing collaborative multi-agent workflows, orchestration, tool calling, memory, RAG, and autonomous task execution using modern LLM frameworks.