How to Track Background Job Progress in React and FastAPI
Track FastAPI background jobs in React with persisted job IDs, status endpoints, safe polling and refresh recovery. Includes a runnable local example.
Read articleWriting
Writing on building useful software.
Articles on engineering decisions, implementation details, and the tradeoffs behind reliable production systems.
All writing
14 articles
Your API Accepts an Idempotency Key. That Does Not Make It Idempotent.
An Idempotency-Key header is only a label. The real guarantee comes from request fingerprints, PostgreSQL constraints, atomic results, and correct retries.
Read articleYour Tenant Context Outlives the Request
How transaction-local tenant context, PostgreSQL Row-Level Security, and connection-pool tests prevent one request from inheriting another tenant's access.
Read articleYour /health Endpoint Is Lying
A 200 from /health proves very little. Useful health checks separate process life, traffic readiness, dependencies, workers, data freshness, and user journeys.
Read articleYour Audit Log Is Not an Audit Trail
An application log can say that an update ran. An audit trail must preserve who changed what, why it changed, and which exact version was affected.
Read articleYour Database Commit Succeeded. Your Message Did Not.
A database transaction and a message provider cannot share one commit. Here is how an outbox table, leased worker, and idempotency close the failure window.
Read articleFour Decisions I Made Before Writing Any Features
Four architecture decisions that kept an enterprise expense system auditable: claim boundaries, immutable history, identity, and concurrent writes.
Read articleYour Agent Still Has the Tool You Revoked
An agent can retain a revoked tool. Why long-running workflows create a TOCTOU authorization bug — and why the decisive guard belongs immediately before provider I/O.
Read articleA Timeout Doesn't Tell You Whether the Write Happened
Retrying an external write is a guess about which of three worlds you are in. Idempotency keys, business-key deduplication, and the database invariants that make the guess unnecessary.
Read articleTracing a Request You're Not Allowed to Log
Correlating one job across TypeScript, a queue, and Python — without turning routine telemetry into a second copy of the sensitive payloads it describes.
Read articleDon't Put a Model Where an If-Statement Will Do
Every model call in a routing path converts a fact into a guess. What I learned about spending probabilistic routing only where the ambiguity is real.
Read articleGenerated Charts Are Untrusted Code
Why a model-generated Mermaid diagram or ECharts spec is a program, not data, and the two-tier validation and bounded repair pipeline I built around it.
Read articleThe Model Wasn't the Hard Part: My Learnings After Shipping Multi-Agent Systems
Two and a half years of building agent systems, and what I got wrong about architecture, context, evaluations, model routing, permissions, and recovery.
Read articleRetrying an AI Agent Is Not the Same as Retrying an API Call
How leases, checkpoints, ownership-safe writes, and tool-history repair made Research-AI recoverable — and what the design still does not guarantee.
Read article