02 / Engineering
Areas of engineering I work in and continue to study.
01
Backend Architecture
Most of my production work sits here: designing service boundaries and APIs that hold up under real traffic, not just in a demo. NestJS and TypeScript are my default toolchain, with PostgreSQL and Redis as the data layer.
Problems
- Designing API boundaries for systems that outgrow their first version
- Migrating an existing Node.js codebase to a structured NestJS architecture
- Query optimization and caching for high-traffic routes
- Auth, rate limiting, and input validation as first-class concerns
Technologies
NestJS, TypeScript, Node.js, PostgreSQL, Redis, REST, GraphQL
02
Distributed Systems
Concurrency and consistency stop being theoretical the moment two bids land on the same auction in the same millisecond. This domain is about designing for that reality — explicit consistency boundaries, not optimistic assumptions.
Problems
- Concurrent writes to shared state without race conditions
- Keeping distributed state consistent under partial failure
- Asynchronous, event-driven processing that stays correct under retries
- Recovering cleanly from a crashed worker mid-transaction
Technologies
PostgreSQL row-level locking, Redis, WebSockets, Message queues
Systems
03
Financial Infrastructure
Payment code is the part of a system that can't be "mostly right." I design for idempotency, webhook reliability, and reconciliation from the first line, not as a fix after a bug report.
Problems
- Buyer deposits, seller payouts, and platform fee splitting on a marketplace
- Webhook-driven reconciliation that survives duplicate or out-of-order delivery
- Refunds, disputes, and failure handling as designed paths, not exceptions
- Keeping payment state and business state from ever disagreeing
Technologies
Stripe, Stripe Connect, PayPal, Webhooks
Systems
04
Cloud & Reliability
Infrastructure decisions are product decisions. I build CI/CD, containerization, and deployment architecture so that shipping fast and staying reliable aren't in tension.
Problems
- Zero-downtime deployments for platforms that can't tolerate a maintenance window
- Reducing cloud cost without degrading reliability
- Monitoring and alerting that catches problems before users do
- Migrating manual, error-prone deployments to automated pipelines
Technologies
AWS, Docker, Kubernetes, GitHub Actions, Terraform, Nginx
05
Intelligent Systems
AI features that survive contact with production, not demos that hallucinate under real data. This means evaluation, observability, and failure handling built in from the start, alongside the model itself.
Problems
- Integrating AI screening/scoring into a workflow with hard latency and reliability requirements
- Isolating AI provider calls so their failures don't take down the rest of the platform
- Evaluating whether a prompt or pipeline change actually improved output quality
- RAG pipelines that don't hallucinate under real, messy source data
Technologies
LangChain, LangGraph, OpenAI API, Claude API, Langfuse
Systems