Amit Verma
Software Engineer · Distributed Systems

Amit Verma

Backend engineer · Kafka, IoT, distributed systems. I write long-form posts the same way I debug: what broke first, what tradeoff you pay, what shipped in production.

At work: 10k+ WebSocket connections, Kafka-scale pipelines, ClickHouse / Mongo / Redis — real constraints, not demos. I also keep DSA sharp with LeetCode (~1,000 problems solved · contest rating ~1,577 (~top 27% globally)).

iot-filtering-problem.js
// Real problem I solved at work
// User has 1 lakh device IDs → filter MongoDB
 
// ❌ Approach 1 — naive $in (100k IDs)
db.device_data.find({
  device_id: { $in: deviceIds }
})
// Timeout. 4.2MB payload. Index dies.
 
// ❌ Approaches 2–6 also failed...
 
// ✅ Approach 7 — store ancestry at ingestion
db.device_data.find({
  "ancestor_nodes.id": {
    $in: ["N2", "N6"] // 2 IDs, not 100k
  },
  temperature: { $gt: 30 }
})
// ✅ 38ms. Works at any scale.
10k+
Concurrent WebSocket connections
50%
P99 latency reduction achieved
~3yr
Production backend (approx.)
1k+
LeetCode problems solved
Rating ~1,577 · ~top 27% globally

Coding practice: @Amitverma2216~1,000 problems solved · contest rating ~1,577 (~top 27% globally). Same discipline as production: ship often, measure, iterate.

01Stack & focus

Tools & focus

~3 years in production backend roles

What I ship with most often. Deeper on Node/TS; honest exposure on the rest.

If your JD mentions Kafka, polyglot databases, AWS, WebSockets, or strong DSA alongside backend work — that’s the lane I fit.

Languages & runtime

Node/TS in production; Go for services, tooling, and performance-sensitive pieces.

  • JavaScript
  • TypeScript
  • Node.js
  • Go

APIs & real-time

REST + sockets under load.

  • Express.js
  • Socket.IO
  • WebSockets

Data & search

Right store for the job.

  • PostgreSQL
  • MySQL
  • MongoDB
  • ClickHouse
  • Redis
  • Elasticsearch
  • Neo4j

Streaming & messaging

Events, fan-out, backpressure.

  • Apache Kafka
  • AWS SQS / SNS
  • RabbitMQ

Cloud & delivery

AWS-first, repeatable deploys.

  • AWS (EC2, RDS, SQS, Lambda, ALB, ASG)
  • Docker
  • GitHub Actions

Observability

Know when things fail.

  • Prometheus
  • Grafana
  • Loki
  • SLA-style alerting

AI in production

Wiring LLMs into real systems — not demos.

  • LLM APIs
  • RAG
  • prompting
  • tool-calling
  • operational chat over telemetry

CS fundamentals & practice

Consistent problem solving — keeps algorithms and data structures sharp next to production work.

  • LeetCode @Amitverma2216
  • ~1,000 problems (Easy / Medium / Hard)
  • Contest rating ~1,577 · ~top 27% globally
02About me
Amit Verma — Software Engineer · Distributed Systems

I build systems at scale
and write about what breaks.

I'm a Software Engineer with about three years building production distributed systems. I work at a Stealth Startup (Gurugram, India), where I architect real-time data pipelines with Kafka, multi-database strategies across MongoDB, ClickHouse, and PostgreSQL, and high-concurrency WebSocket infrastructure for IoT and NOC-style platforms — including work that serves enterprise deployments for clients such as Reliance and Altius Infra.

Previously I shipped notification systems handling 1M+ daily notifications at JP Furnware & Technologies, and AI-powered SaaS at PinnacleWorks Infotech. I hold a B.Tech in Computer Science from CCS University (CGPA 8.29/10).

I keep data structures & algorithms current with steady LeetCode practice (~1,000 problems solved · contest rating ~1,577 (~top 27% globally)) — it rounds out production engineering with measurable problem-solving depth.

This site exists because most engineering content shows you the clean answer. I show you the seven approaches that failed first — the real thinking, the edge cases, and the honest tradeoffs you have to make at scale.

Software Engineer
Stealth Startup
Jan 2025 – Present · Gurugram
Kafka · WebSockets (10k+ connections) · ClickHouse · Multi-tenant IoT / NOC SaaS · LLM integration · AWS · Grafana/Prometheus · ~50% P99 latency reduction
Backend Developer
JP Furnware & Technologies
Jan 2024 – Jan 2025 · Gurugram
AWS SQS · WebSockets · Firebase · 1M+ daily notifications · 900ms → 40ms query optimization
Software Developer
PinnacleWorks Infotech
Jul 2023 – Dec 2023 · Gurugram
NLU engines · Node.js EventEmitter · Redis caching · 40% fewer redundant calls

Also in the mix when the product needs it: Prisma, React, Airflow — not the headline, but real shipping work.

03Contact

Get in touch

Hiring, collaboration, speaking, or a question about something I wrote — send an email. I read everything.

amit.verma.codes@gmail.com

Uses your default mail app (Gmail, Outlook, Apple Mail, etc.) — no form, no tracking.

Latest writing

AdvancedFeatured
001

The Brutal Truth About Multi-Tenant IoT Data Filtering: Every Obvious Solution Failed

A user had access to 1 lakh devices across MongoDB, ClickHouse, and Elasticsearch. We tried seven approaches. All broke. The full journey — dead ends, edge cases, and what production systems actually do.

System DesignIoTDistributed Systems·15 min
Medium
002

Vector Search Explained: From KNN to HNSW — A Complete Guide

How similarity search works under the hood — and why naive brute-force collapses at scale.

SearchML·13 min
Medium
003

MySQL Query Optimization: Real Lessons from a 36GB Index Disaster

What happens when your index outgrows your data — and how we fixed it without downtime in production.

MySQLProduction·5 min
Beginner
004

How Twitter Generates Your Home Feed in 5 Seconds

Fan-out on write vs fan-out on read — and why Twitter uses a hybrid at 500M users. A beginner-friendly breakdown.

System Design·8 min
Advanced
005

Socket Exhaustion in High-Concurrency IoT Systems

A production case study — how a small config issue brought down an entire IoT device fleet and what fixed it.

IoTNode.js·4 min

More essays on Medium — same voice: failures first, tradeoffs explicit, production-shaped problems only.