Keyboard shortcuts

1
Jump to Home
2
Jump to About
3
Jump to Tech stack
4
Jump to Architecture
5
Jump to Projects
6
Jump to Experience
7
Jump to Contact
?
Show this help
Esc
Close dialogs & panels

psst — every box and arrow on this board is drawn live with rough.js ✎

Hi, I'm

Noor

Backend & AI Engineer

I design and build scalable backends and AI systems that stay fast, observable and boring under load.

See my work →GitHubSay hi

currently

Scaling RAG pipelines & event-driven services

open to

Backend / AI engineering roles

scroll ↓

Frame 2 · about

About me

I'm a software engineer focused on the unglamorous parts that make products work: APIs, data pipelines, queues, caches and the infrastructure that keeps them running.

Lately I've been building AI-powered services — retrieval pipelines, LLM orchestration and evaluation — with the same bar for reliability as any other production system.

How I work

  • Design for the 10x traffic day
  • Measure before optimising
  • Make failure modes explicit
  • Ship small, observe, iterate

  • 2+

    years shipping backends

  • 10M+

    requests / day served

  • 15+

    services in production

Frame 3 · stack

Tech stack

The tools I reach for — though the problem picks the tool, not the other way round.

  • Languages

    • Javascript
    • Python
    • Go
    • TypeScript
    • SQL

  • Backend

    • FastAPI
    • Node.js
    • gRPC
    • REST
    • GraphQL

  • Data & Queues

    • PostgreSQL
    • Redis
    • Kafka
    • RabbitMQ
    • MongoDB

  • Cloud & DevOps

    • AWS
    • Docker
    • Kubernetes
    • Terraform
    • GitHub Actions

  • AI / ML

    • LLM APIs
    • RAG
    • LangGraph
    • pgvector
    • Evals
  • + whatever the problem needs next

Frame 4 · architecture

How I design systems

A simplified sketch of Realtime analytics + RAG platform — the kind of system I build: async where it can be, cached where it counts, observable everywhere.

eventsconsumepersistembedcachemissRAGcontextClientsweb · mobileAPI Gatewayauth · rate limitIngest svcGoQuery svcFastAPIAuth svcOIDCKafkaevent streamRediscacheLLMgenerationWorkersconsumersPostgressource of truthVector DBpgvector

↑ hover or tap any box to see what it does

  1. 1

    Every request passes the gateway: auth, rate limits and tracing happen once, at the edge.

  2. 2

    Writes go async through Kafka, so a traffic spike queues up instead of taking the API down.

  3. 3

    Reads hit Redis first and fall back to Postgres, keeping p99 latency low.

  4. 4

    Questions are answered with RAG: retrieve from the vector DB, generate with the LLM, stream back with citations.

  • 50k/s

    peak events

  • <120ms

    p99 read

  • 99.95%

    availability

Frame 5 · projects

Selected projects

Click a card to open its details — the problem, what I built, and what changed.

Frame 6 · experience

Experience

  1. 2026 — now

    Backend / AI Engineer

    Company Name

    • Own the RAG platform and LLM services used by internal teams
    • Cut p99 latency of the core API by 45% via caching and query tuning

  2. 2025 Jul — 2026 Mar

    Full-Stack Engineer

    Trackepay Pvt Ltd

    • Built the event ingestion pipeline processing 50k events/s
    • Led migration from a monolith to independently deployed services

Frame 7 · contact

Let's build something

Have a system that needs to scale, or an AI feature that needs to be production-grade? I'd love to hear about it.

Leave a note