AI Product Builder · Backend Engineering

Muhammad
Waris
Ali.

I build AI-powered software products and the backend systems behind them.

Founder of Flintel, a research tool that finds and classifies buyer-intent conversations on Reddit and Hacker News. Builder of Writora, a free AI writing assistant. I work mostly in Python and FastAPI, with MongoDB and LLM APIs.

See Flintel → Projects Resume ↓
Portrait of Muhammad Waris Ali
Muhammad Waris Ali · Pakistan
About
What I Build With

I'm a software developer from Pakistan building AI-powered products and the backend systems behind them. I founded Flintel and built Writora, and most of my work is designing the data model, wiring in LLM calls, and shipping the result as something people can use. I'm also working through a Backend AI Engineering internship track at FlyRank AI.

01 — Backend

Backend Engineering

Python FastAPI asyncio REST API design WebSockets Background workers JWT & OAuth 2.0
02 — AI

AI Integration & Applications

Claude API OpenAI API Function calling / tool use Prompt engineering LLM-based classification Conversational assistants
03 — Data

Databases & Data Architecture

MongoDB Atlas Motor (async) Schema design & indexing Multi-tenant isolation Caching Crash-safe batch processing
04 — Integrations

APIs & Integrations

WhatsApp Business API Duffel ClipDrop Paddle Billing Resend Webhooks
05 — Infrastructure

Deployment & Infrastructure

Render Git & GitHub
Live Product · Founder & Builder

Flintel —
Social Research
Reddit & Hacker News

People often say what they need before they ever talk to a vendor: "alternatives to…", "what do you use for…". Those posts are scattered and slow to triage. Flintel lets you search a brand, product, or topic in plain language and returns the matching conversations from Reddit and Hacker News, each linked to its source post and labelled by intent. It is built for founders, marketers, and researchers who want to read what people are actually saying before they act.

  • Each result carries a summary, source, date, sentiment, and an intent label (High, Medium, Low, or Uncertain) with a short reason
  • A strict definition of "qualified": High requires an explicit purchase step, and rows that can't be verified against the stored post are shown as Uncertain rather than inflated
  • Backend I build and maintain: background discovery and collection workers, LLM analysis with structured output, and a shared plus per-user MongoDB data layer
  • Free and paid plans with monthly search credits. Reddit and Hacker News are the live sources today; X, LinkedIn, and Facebook are listed as coming soon, and Flintel does not contact anyone on your behalf
Visit Flintel → How intent labels work
flintel.status Live
Reddit & Hacker News search

Plain-language search over collected posts, with a summary and a link to every source post.

StatusAvailable
Intent labels

High, Medium, Low, or Uncertain, each with a reason and, where one exists, an excerpt from the post.

StatusAvailable
X, LinkedIn, Facebook

Shown as "coming soon" in the app. Not part of the product today.

StatusPlanned
2
Live Sources
4
Intent Labels
3
Platforms Planned
Projects
Selected Work
✍️
Product Live Builder

Writora — Free AI Writing Assistant

Most people know what they want to say but struggle to phrase it in emails, replies, and messages. Writora takes rough text and returns a clearer version, with no account required. It is aimed at students, professionals, and non-native English speakers.

  • Six tools: improve writing, write a reply, change tone (14 options), simplify, shorten, and fix grammar
  • Dedicated pages for common tasks such as professional emails, follow-ups, and polite refusals
  • Each tool runs on its own task-specific instructions; text is sent to a third-party AI service and, per its privacy page, not stored
  • English works best today; other languages are listed as coming soon
Visit Writora → About
📱
Demo FastAPI · WhatsApp API

WhatsApp Restaurant Assistant

Customers place orders and book tables through WhatsApp chat in English, Urdu, and German. Staff manage everything from an admin CRM.

  • Multi-item order parsing with fuzzy product matching
  • Natural-language reservation booking and delivery-charge logic
  • LLM fallback via function calling for open-ended questions
  • FastAPI, MongoDB, and JWT-protected admin CRM
CRM Dashboard → Demo post
🛒
Demo FastAPI · WebSockets

AI Shopping Assistant

A multilingual WebSocket shopping assistant that turns natural-language requests into product searches and ranks the results.

  • English, Urdu, and German with language detection
  • Keyword and intent matching, relevance scoring, and natural-language price parsing
  • Analytics on searches, clicks, and language usage
  • MongoDB Atlas persistence
Live App → Demo post
✈️
Demo FastAPI · MongoDB

Travelfy — AI Travel Assistant

A conversational assistant that plans trips through chat and keeps context across turns.

  • Multi-turn trip planning with per-session persistence
  • MongoDB Atlas on Render
Live App →
🖼️
Live FastAPI · ClipDrop

PixUp8K — Image Upscaling Backend

PixUp8K is a SaaS for upscaling photos to 4K/8K and removing backgrounds. I built and maintain its backend.

  • REST API with JWT auth and per-feature usage quotas
  • WebSocket notifications for processing status
  • Image history tracking; ClipDrop API and Pillow for processing
Visit PixUp8K →
System Design
Flintel
Architecture

Flintel is split into two services: a background engine that discovers and collects posts, and the user-facing web app that searches and presents them. Keeping collection and analysis out of the request path makes searches fast and lets long batch jobs recover from restarts.

01 — Collection

Discovery Engine

Multi-threaded background workers collect Reddit posts through Google SERP discovery and auto-generated fuzzy keyword matching.

02 — Analysis

LLM Intent Labelling

Claude/GPT scoring with function calling for structured output. Rows that can't be matched to a stored post are labelled Uncertain.

03 — Data

Shared + Per-User Cache

Async MongoDB (Motor) with a shared cache and per-user data. Batch processing is crash-safe and restartable.

04 — Product

Web App

FastAPI web app with plain-language search, source-linked results, search history, and credit-based plans.

Experience
Experience
Internship 01.07.2026 to 09.07.2026

Backend AI Engineering Intern — FlyRank AI

Backend AI Engineering Internship Program at FlyRank AI.

  • Backend AI engineering assignments and capstone work in Python, using REST API design, within FlyRank AI's Backend AI Engineering track
  • Applied LLM integration, including function calling and tool-use patterns, along with backend architecture and system design practices, to project briefs
See Internship Post →
Credentials
Certifications
View Certificate Post →
Backend & Chatbot Development
01

Chatbot Development Guide 2025 — Beginner + Advanced

Udemy · Eric Roby · Sept 2025 · 10 Hours · Certificate

Completed
02

Backend AI Engineering

FlyRank AI · Internship Program · 01.07.2026 to 09.07.2026 · Issued 05.11.2026 · ID FR-D11-80064-54835 · Verify

FlyRank AI Certificate of Completion: Backend AI Engineering, issued to Waris Ali Completed
03

AI / Python Backend Development

Udemy · Completed 2025

Completed
04

API Development & FastAPI Training

Online Certification · Completed 2025

Completed
LET'S
BUILD.

Open to conversations about AI products, backend systems, and collaboration.

Start a Conversation → LinkedIn Resume ↓
✉ abdulwaarisinfo@gmail.com