faizan

Hi, I'm Faizan

Backend & Applied AI Systems Engineer

I build high-throughput distributed backends, event-driven messaging pipelines, and production AI agents. Currently scaling backend infrastructure, OAuth-secured MCP servers, and multi-tenant RAG systems at FutureBlink.

Work Experience

FutureBlink logo

FutureBlink

June 2024 - Present

Software Engineer

Led backend scalability initiatives for by implementing distributed locking via atomic MongoDB operations with stale-lock recovery, enabling horizontal scaling across microservices while eliminating race conditions and duplicate email dispatches.
Drove API performance optimizations by eliminating N+1 queries via batched set-based database lookups, adding targeted ESR compound indexes, and implementing a bounded in-memory FIFO cache, boosting throughput by up to 5x and slashing response times by 90%.
Architected and deployed an OAuth-secured Model Context Protocol (MCP) server with strict transport-level token scoping, enabling external AI clients and internal automated support agents to safely query real-time user context and execute scoped actions.
Re-architected the email response tracking engine by introducing OAuth mailbox push webhooks, RabbitMQ message queues, and time-and-size bounded consumer micro-batching alongside tiered SMTP/IMAP polling, cutting reply detection latency from hours to seconds.
Engineered a multi-tenant, RAG-based AI voice agent for Shopify merchants using Vapi, Twilio, LangChain, and Pinecone (with per-merchant namespace isolation), delivering low-latency, context-aware customer support.
TypeScript
Node.js
MongoDB
RabbitMQ
Model Context Protocol (MCP)
PostgreSQL
Redis
Pinecone
LangChain
Vapi
Twilio
Express
Next.js
Docker

Case Studies

Distributed Systems
Hours → Seconds

Event-Driven Email Sync Architecture

Re-architected a legacy polling engine into an event-driven pipeline using OAuth mailbox push notifications, RabbitMQ message queues, and time-and-size bounded consumer micro-batching, cutting reply detection latency from hours to seconds.

Core Architectural Patterns

  • Time-and-count bounded micro-batching (10–60s window or size thresholds) to eliminate redundant inbox connections.
  • RabbitMQ manual ACK confirmations ensuring zero event loss during worker crash or failover.
  • Tiered priority queues separating high-throughput OAuth webhooks from legacy polling mailboxes.
RabbitMQ
OAuth Webhooks
IMAP/SMTP
AI Infrastructure
OAuth 2.0 Scoped

OAuth-Secured Model Context Protocol (MCP) Server

Architected and deployed a remote MCP server with OAuth 2.0 token scoping, enabling external AI clients and internal automated support agents to safely query real-time user context and execute actions.

Core Architectural Patterns

  • Transport-level OAuth 2.0 token enforcement strictly preventing tenant cross-talk or prompt injection leaks.
  • Unified tool schema abstraction allowing both internal support bots and external LLM clients to discover and invoke endpoints safely.
  • Granular permission scoping mapping customer roles directly to executable backend tool actions.
TypeScript
MCP
OAuth 2.0
Applied AI & Voice
Namespace Isolated

Multi-Tenant E-Commerce Voice RAG Agent

Engineered a real-time conversational voice agent for Shopify merchants using telephony orchestration and namespace-isolated vector search to deliver low-latency, context-aware customer support.

Core Architectural Patterns

  • Deterministic Pinecone namespace isolation ensuring 100% tenant boundary security and targeted vector search.
  • Compact chunking strategy with streaming token integration to maintain sub-second voice conversational cadence.
  • Event-driven vector invalidation and upsert pipeline synchronized with merchant catalog and policy updates.
Vapi
Twilio
LangChain
Pinecone

Projects

Quiz Master

Quiz Master

A dynamic, real-time multiplayer web application that allows users to join or host engaging quiz rooms. Participants can compete by answering AI-generated questions on a variety of topics, providing an interactive and competitive experience.

Google Gemini API
Cloudflare Workers
React
Express
PostgreSQL
Prisma
WebSockets
Code Preview
Wings In Progress

Wings In Progress

A comprehensive application that provides aspiring pilots a platform to track their flight training progress. It simplifies essential tasks like logging flights, tracking progress, and analyzing performance metrics through an intuitive user interface.

Next.js
MongoDB
Mongoose
React
React Query
TypeScript
NextAuth.js
Zod
Code Preview

Skills

Backend & Distributed Systems

Node.js

Microservices

RabbitMQ

Distributed Locking

REST APIs

WebSockets

OAuth 2.0

Stripe

AI & Agent Infrastructure

Model Context Protocol (MCP)

RAG

LangChain

Pinecone

Vapi

Twilio

Google Gemini API

Databases & Languages

TypeScript

Python

C++

MongoDB

PostgreSQL

Redis

Frameworks & DevOps

Express

Next.js

Django

Prisma

Mongoose

React

Docker

Git

Let's build something great!

Open to discussing high-throughput distributed backends, event-driven messaging pipelines, and applied AI systems opportunities. Feel free to reach out to connect or discuss your team's engineering challenges.