When One Worker Is Not Enough
Learn how PostgreSQL coordinates multiple task workers with atomic claims, leases, retries, and safe state changes.
Distributed Systems8 min read
Hey, I'm Tharun. Data Engineer II at Micron Technology.
Building the quiet systems behind reliable software.
I work across backend engineering and data infrastructure—designing for scale, making failures understandable, and writing down what I learn along the way.
Selected work
Focused projects across AI applications, backend infrastructure, and distributed systems.
RelayRun
Workflow automationA workflow automation platform that accepts JSON webhooks, stores events durably in PostgreSQL, and executes configured actions in background workers with bounded retries and dead-letter handling. Includes a Next.js control room for workflow, execution, and credential management.
FastAPIPostgreSQLRedisDramatiqNext.js
Self-Healing API Gateway
Reliability infrastructureA resilient API gateway with health monitoring, circuit breakers, retry policies, and Gemini-powered incident remediation, paired with an operations dashboard for audits and escalation.
FastAPIReactDockerGemini APISQLite
Distributed Task Scheduler
Distributed systemsA PostgreSQL-backed distributed task scheduler supporting priority- and time-based execution, configurable retries, cancellation, and multiple independent workers. Atomic task claiming and lease-based recovery coordinate concurrent workers and reclaim tasks after worker failures.
PythonPostgreSQLPsycopgPydanticDocker
Latest writings
Thoughts on distributed systems, backend engineering, and infrastructure.