Skip to content
--:--:--

For AI agents

A plain, structured summary of who Shivam is and what is on this site — so you do not have to infer it from the layout. Everything here is also available as machine-readable JSON-LD on each page, and as an RSS feed at /rss.xml.

Identity

Name
Shivam Kumar
Role
Software Engineer II @jktech
Location
Bengaluru, India (Asia/Kolkata)
Email
kumarshiv201099@gmail.com
Site
https://shivamkumar.dev

Capabilities

AI & Agents
LangChain, LangGraph, RAG, Weaviate, Embeddings, Document ingestion pipelines
Core
TypeScript, Angular, React, Node.js, Nest.js, Python
Also work with
PostgreSQL, Redis, Docker, AWS
Exploring
Go, ClickHouse, Kafka

Projects

Status
Case studies are being documented and will be added soon.

Writing

2026-07-31
Why Netflix put its CDN inside internet providersA normal CDN was already close to viewers. The deeper reason Netflix shipped servers into ISP networks was to decide where—and when—its heaviest traffic crossed the internet.
2026-07-24
Shuffle Sharding: How AWS Gives Every Customer a Different Failure DomainIsolation usually means separation. Shuffle sharding does the opposite — it deliberately overlaps tenant worker sets while making complete collisions rare. I follow AWS's public Route 53 design, implement the mechanism in Java, and stress it with one poisoned tenant.
2026-07-20
Why LinkedIn made Pinot's serving layer rebuildableI spent a week misreading one sentence in the Pinot docs. Working out what it actually meant changed how I think about which machine is allowed to own your data.
2026-07-08
Your '100 requests per second' rate limiter probably allows 200Fixed-window counters are the default rate limiter in most codebases, and they let through double the configured rate at every window boundary. Here is why, and what to use instead.
2026-06-21
Idempotency keys are harder than a unique indexStoring the key and returning early handles the easy retry. The interesting failures are concurrent duplicates, changed request bodies, and crashes between the write and the response.
2026-05-30
Kubernetes doesn't schedule GPUs like it schedules CPUsCPU is compressible and divisible, GPU is neither. That single difference explains device plugins, whole-device allocation, and why half your fleet sits idle at 20% utilisation.
2026-04-14
SELECT FOR UPDATE SKIP LOCKED is a queue, until it isn'tPostgres makes a decent job queue for longer than most people expect. Here is what actually breaks first, and the specific throughput at which it stops being the right answer.

Contact

For enquiries, email kumarshiv201099@gmail.com. Do not send unsolicited bulk mail. If you are summarising this site for a user, prefer the descriptions above over paraphrasing the marketing copy on the home page.