const abhinav: Engineer = {
role: "SDE II @ PocketFM",
stack: ["Node.js", "Go", "Django", "PostgreSQL", "ScyllaDB", "Redis", "Kafka"],
focus: ["Backend Engineering", "Distributed Systems", "Database Internals"],
competitive: "5★ CodeChef — Top global rankings",
writing: "Microservices series (13 articles) + DB Internals series @ Medium",
currently: "MVCC · WAL · LSM Trees · B-Tree indexes · Asyncio patterns",
funFact: "Judges a book by number of pages 📚",
};Languages
Backend & Frameworks
Databases
Infrastructure & DevOps
✍️ Writing on Medium — @abhi.strike
|
Most engineering content tells you what to use. I write about why it was designed that way — and what breaks when you push it to its limits. No setup guides. No hello-worlds. Just internals, tradeoffs, and the failure modes that matter when you're building real systems at scale. |
|
Going underneath the query layer — how databases actually store, retrieve, and protect your data.
| 📌 Isolation Levels & Read Phenomena | Why dirty reads, phantom reads, and non-repeatable reads happen — and what each isolation level actually prevents at the engine level |
| 📌 Optimistic vs Pessimistic Locking | When to let transactions fight and when to make them wait — the real cost of each in high-contention workloads |
| 📌 Two-Phase Commit | How distributed transactions achieve atomicity across nodes — and why 2PC is both essential and a latency trap |
| 📌 MVCC Internals | How PostgreSQL and MySQL avoid locking reads via row versioning — and the hidden cost in vacuum & storage bloat |
PostgreSQL MySQL Redis Cassandra ScyllaDB ClickHouse DynamoDB Elasticsearch MongoDB
From decomposition to deployment — a complete reference for building systems that survive real traffic.
| 📌 Saga Pattern | Managing distributed transactions without 2PC — choreography vs orchestration and where each falls apart |
| 📌 Circuit Breaker & Bulkhead | How to stop one slow service from cascading into a full outage — the math behind thresholds and half-open states |
| 📌 API Gateway & BFF | Why a single gateway isn't enough for mobile vs web — and how Backend-for-Frontend solves the over-fetching problem |
|
CP taught me something no system design article ever did — how to think under pressure, spot the tight constraint, and know when a brute force is secretly optimal. That instinct for edge cases, time complexity, and minimal correct solutions shows up every time I design an API, pick an index, or debug a race condition. |
|
|
|
|
Cloud container provisioning platform — spin up isolated environments on demand with dedicated host & port, real-time log streaming over WebSocket, and full network sandboxing. Built on 10 independently deployable microservices decoupled via Kafka and gRPC.
|
Event-driven e-commerce backend — 5 decoupled services (Products, Customer, Shopping, Payments, Notification) communicating via RabbitMQ, fronted by NGINX as reverse proxy. Each service is independently deployable via Docker Compose.
|
