Muhammad Zain Naseer

Full-Stack Software Engineer at INDUS Technologies | React, TypeScript, Node.js, FastAPI, PostgreSQL | I build full-stack web apps and backends that sync 10,000+ companies a day

Based in: Lahore, PakistanCurrently: Junior Fullstack Engineer, INDUS TechnologiesExperience: 2 years

Hire me

The software I work on tracks every worker, machine and material on WAPDA's construction projects. At INDUS Technologies I wrote Atom Manager, the part of the DipGos platform that knows where each resource is and what it has done so far. I work across the stack. Most of my front ends are React and TypeScript, and the back ends are FastAPI or Node.js on PostgreSQL. Before INDUS I spent ten months at Attributter on the backend that connects LinkedIn Ads to HubSpot. The hard part was syncing more than 10,000 companies every day inside the API rate limits. I solved it with a Redis/BullMQ job queue and a token bucket rate limiter. At LUMS I led development of SensAI, which listens to a doctor's consultation and writes the clinical notes. With three classmates I built HerSafar, a ride-sharing app that women can use by speaking Urdu instead of reading and typing. I like building automations too. One of mine approves WhatsApp join requests for LUMS Opportunities, a student community of 6,000+ members, so its admins no longer check each request by hand. I've taught as well: two semesters as a teaching assistant for CS-100 at LUMS, and three groups of Federal Board of Revenue officers learning Python, SQL and Excel. My projects are at zainnaseer.com.

Experience

Junior Fullstack Engineer

INDUS TechnologiesLahore, Pakistan (On-site)workJun 2026 – Present

Summary

Full-stack engineer on DipGos (Digital Project Governance and Operations System), a project governance platform deployed to WAPDA (Pakistan Water & Power Development Authority).

What I did

Built the Atom Manager business function end to end — React/TypeScript frontend and FastAPI/PostgreSQL backend.
Designed the resource domain model spanning workforce and material resources: kinds, atoms, units, lots, vessels and holdings.
Built a multi-step resource creation wizard with per-step validation and duplicate-tag detection, and scoped the resource catalogue across every level of the project hierarchy.
Work on DipGos / Atom Manager for WAPDA using FastAPI, PostgreSQL, and React/TypeScript.

Results

DipGos is in use at WAPDA (Pakistan Water & Power Development Authority).

ReactTypeScriptFastAPIPythonPostgreSQLSQLAlchemy

Teaching Assistant — Business Analytics & Data Science (FBR)

Rausing Executive Development Centre (REDC), SDSB-LUMSLahore, Pakistan (On-site)part timeFeb 2026 – May 2026

Summary

Selected as Teaching Assistant across three consecutive 3-week intensive certification programmes for Grade 17–18 officers of the Federal Board of Revenue.

What I did

Supported instruction across 15 sessions per week covering Excel, Python and SQL for senior government professionals.
Assisted with lab sessions, graded assignments, and provided hands-on one-to-one guidance on data analytics concepts and tools.
The May 2026 cohort covered Excel modelling, financial statement analysis, ratio analysis and revenue forecasting, including Excel-based forecasting tools and pivot tables.

Results

Three cohorts of Federal Board of Revenue officers

Microsoft ExcelPythonSQLData AnalyticsTeaching

Back End Developer (Part-time)

Attributter (Market Pro)Lahore, Pakistan (Remote)part timeJul 2025 – Apr 2026

companies synced every day: 10,000+companies synced every day

Summary

Architected backend infrastructure connecting LinkedIn Ads and HubSpot CRM using Node.js, Express.js, MongoDB and PostgreSQL, with automated testing and error handling.

What I did

Engineered a Redis/BullMQ job orchestration system for non-blocking daily sync of 10,000+ companies across multiple users via parent-child API chaining (50 API calls per sync cycle).
Implemented a token bucket rate limiter, JWT authentication and data validation middleware for production-ready REST APIs.
Built CI/CD pipelines using Docker and GitHub Actions, deploying to Hetzner.

Results

The sync runs every day without blocking and stays inside the LinkedIn and HubSpot API rate limits.

Node.jsExpress.jsMongoDBPostgreSQLRedisBullMQJWTDockerGitHub ActionsHetzner

Skills

Technical

JavaScript

TypeScript

Python

C++

SQL

Node.js

Express.js

GraphQL

Flask

Socket.IO

BullMQ

React.js

Flutter

PostgreSQL

MongoDB

Redis

Docker

Kubernetes

AWS

Hetzner

Git

GitHub Actions

PyTorch

TensorFlow

OpenCV

HuggingFace

FastAPI

Logistic Regression

Neural Networks

Linear Regression

Prompt Design

Model Evaluation

REST API Design

Data Modeling

Authentication

React

Multinomial Naive Bayes

Unity

SQLAlchemy

Figma

Microsoft Excel

Areas of expertise

Backend Architecture

API Design

Job Orchestration

AI Integration

Real-Time Systems

Computer Vision

Full-Stack Development

Business Process Automation

UX Research

Teaching & Technical Training

Projects

LUMS Opportunities: Sign-Up and WhatsApp Approval System

personal

Sign-up site and WhatsApp approval bot for a student community of 6,000+ members. Students verify their LUMS email with a one-time code and register their number, and the bot approves eligible join requests on its own.

Node.jsExpressMongoDBBaileys

Visit the site

HerSafar — Voice-First Ride-Sharing Platform

Backend Developer, also managed deploymentpersonalAug 2025 – May 2026

Voice-first mobile application for low-literacy female commuters. Integrates Google Gemini to convert Urdu voice input into structured ride-request data. Team of four. I worked mostly on the Node.js backend and managed deployment on Oracle Kubernetes Engine with GitHub Actions CI/CD, and SonarQube for code quality. Safety features include SOS alerts, buddy pairing and trusted contact sharing. Multimodal interaction and real-time ride matching, validated through iterative user research with target commuters.

FlutterNode.jsMongoDBGoogle GeminiOracle Kubernetes EngineGoogle Cloud

WhatsApp AI Assistant

personal

A personal assistant on WhatsApp that remembers each conversation and picks its own tools: web search, page scraping, Google Calendar and Google Sheets.

n8nOpenAIWhatsApp

Structure-from-Motion Pipeline

personalNov 2025 – Dec 2025

3D reconstruction from 2D photographs using camera calibration, feature matching across views, epipolar geometry and bundle adjustment, producing point clouds suitable for interactive virtual tours.

PythonOpenCVNumPyComputer Vision

View the code

AI Lead Follow-Up Automation

personal

Every three minutes it reads new leads from Follow Up Boss, checks which contact details are valid, and writes a tailored email or SMS for each one, sent through Gmail or Twilio.

n8nJavaScriptVertex AIGmailTwilio

Top-View Scene Mapping

personalOct 2025 – Oct 2025

Spatial intelligence pipeline converting ground-level camera views into top-down scene maps using homography transformation combined with semantic segmentation.

PythonOpenCVSemantic Segmentation

Darbaar — Real-Time Multiplayer Social Deduction Game

Group ProjectpersonalMay 2025 – May 2025

Real-time multiplayer social deduction game inspired by the desi game "Badshah Ka Wazir Kon?". Decentralised state management broadcasting individual player state via Socket.IO to keep concurrent game rooms in sync. Full game-room lifecycle: creation, join, role assignment, voting, public chat, timers and resolution.

ReactNode.jsSocket.IOMongoDBVercel

Visit the site View the code

UstaadOnline — Online Tutoring Platform

Backend Developer, also managed deploymentpersonalFeb 2025 – May 2025

Tutoring platform with JWT and Google sign-in, role-based access, tutor-student matching, session booking and Stripe payments. Team of five. I worked mostly on the backend (REST APIs in Node.js and Express on PostgreSQL) and managed the deployment on Vercel. The team tracked work in Jira.

PostgreSQLExpressReactNode.js

Visit the site View the code

Tandurust Pakistan — Visual-First Healthcare Platform

Advisor: Dr. Suleman Shahid, LUMSpersonal2025 – May 2025

Tandurust ("healthy") Pakistan is a mobile health app design for patients who cannot easily read medical terms or text-heavy screens, built by a team of five. Visual-first screens in Urdu, English and regional languages, built as a Figma prototype. Tested with 50+ people across three government hospitals in Lahore over several rounds, and changed the design after each one.

FigmaUX ResearchMultilingual Design

Visit the site

Urdu News Classifier

Group ProjectpersonalNov 2024 – Dec 2024

Scraped 1,000+ Urdu news articles from Geo, Express and Jang using BeautifulSoup to build a labelled corpus across five categories. Implemented three classifiers from scratch — a PyTorch neural network, one-vs-all logistic regression, and multinomial Naive Bayes. Accuracy on the test set: Naive Bayes 97.6%, neural network 95.2% and logistic regression 94%.

PythonPyTorchBeautifulSoupTF-IDFNLP

View the code

2D-Shooting Game

personalMay 2024 – May 2024

2D shooting game built in Unity with sprite handling, collision detection and audio integration.

UnityC#Game Development

View the code

DHT FileVault — Distributed Hash Table System

personalApr 2024 – May 2024

Distributed hash table system engineered from scratch in Python, implementing consistent hashing for key distribution, node replication for redundancy, and fault tolerance for graceful handling of node failures.

PythonDistributed SystemsConsistent Hashing

Education

Bachelor of Science, Computer Science

Lahore University of Management Sciences (LUMS)Computer ScienceGPA 3.2 / 4Sep 2022 – Jul 2026

Coursework

Software EngineeringMachine LearningArtificial IntelligenceComputer VisionDistributed SystemsData Structures & AlgorithmsDatabasesOperating Systems

Pre-Engineering (FSc)

Cadet College Hasan AbdalPre-EngineeringGPA 91 / 100Apr 2017 – Jun 2022

Recognition

Certifications

Complete Computer Vision Bootcamp With PyTorch & Tensorflow

Udemy

Unity tutorial from scratch till infinity

UdemyCredential UC-fc9215ea-6fdd-4564-9544-f5e8b3a1b196

Numpy For Data Science - Real Time Exercises

UdemyCredential UC-1ff80d90-c12e-4ea9-a23c-2333403bc9ac

What drives the work

Motivations: Leveraging AI for social good
Architecting high-scale distributed systems
Designing for accessibility and low-literacy users
Principles: Empathy-driven design
Technical rigor
Inclusivity

Hire MuhammadReach out about a role, a contract or a conversation.For recruiters

Muhammad's twin is AI, it can make mistakes.

More people on Kanvis

Browse more people Create your own Kanvis