Open to Software / AI Engineer roles

Muhammad Asim

Software Engineer at VentureDive, working on retrieval pipelines and AI agents. Before that I spent a year and a half building client software under an experienced developer's review — which is where I learned that shipping is the easy half.

Four projects are written up here in full: a statutory e-invoicing platform, a restaurant operating system, an auction agent that bids within a ceiling it is given, and a campus marketplace. Each write-up covers the constraint, the architecture and the trade-offs I took deliberately.

Read the case studies Résumé

lib/rag/answer.ts

years building: 2years buildingclient work + industry
apps deployed: 3apps deployedmohasib · chefos · intellibid
case studies: 4case studieswritten up in full
cgpa / 4.0: 3.61cgpa / 4.0bs software eng · ucp

01 — about

I build the parts that have to keep working.

I'm a Software Engineer at VentureDive and an AI Engineer, with a BS Software Engineering and a 3.61 CGPA from University of Central Punjab (UCP).

Most of my work sits at the seam between AI and ordinary product engineering: retrieval pipelines that have to return the right document rather than a plausible one, agents that take real actions, and the unglamorous auth, validation and deployment work that decides whether any of it survives contact with users.

Before VentureDive I spent about a year and a half building client software under an experienced developer who owned the client relationships and reviewed everything I wrote. That arrangement taught me more than the code did — how to read a vague requirement, when to push back on scope, and what “done” actually means when someone else is paying for it.

I'd rather write one system carefully than six demos quickly. The three case studies on this site are the honest version of what that looks like: what the constraint was, what I chose, and what it cost.

Currently: Software Engineer Intern, VentureDive — Lahore
Focus: RAG pipelines, AI agents, full-stack product engineering
Education: BS Software Engineering, University of Central Punjab (UCP) — 3.61 CGPA
Based in: Lahore, Pakistan
Open to: Remote worldwide · on-site Lahore · full-time or contract

Stack

AI engineering: RAG pipelines
Autonomous agents
LLM APIs
Vector search
Prompt design
Python
Frontend: React
Next.js
TypeScript
Tailwind CSS
React Three Fiber
Accessibility
Backend: Node.js
Express
REST API design
Socket.io
JWT & OAuth
Webhooks
Data: MongoDB
Mongoose
Redis
MySQL
Indexing
Schema design
Delivery & cloud: Docker
GitHub Actions
CI/CD
AWS
Nginx
Git Flow
Integrations: Stripe
Safepay
PRAL / FBR API
Cloudinary
Google Generative AI
WhatsApp API

02 — experience

Where I've worked.

Two roles, listed with what I actually owned rather than what the team did.

Feb 2026 — Present
Lahore, PK
current
Software Engineer Intern
VentureDive
Shipping AI-powered features inside production codebases alongside senior engineers.
- Build and ship full-stack features in production Next.js/Node services, from ticket to code review to release.
- Implement retrieval-augmented generation (RAG) pipelines — chunking, embedding, vector search and grounded prompting — so LLM answers cite real source documents instead of hallucinating.
- Work inside CI/CD workflows: automated checks on every pull request, reviewed commits, and staged deploys.
- Contribute to system-design discussions on data modelling, service boundaries and API contracts.
- RAG
- System Design
- CI/CD
- Next.js
- Node.js
- Code Review
1.5 years
Remote, Lahore
Full-Stack & AI Developer — client projects
Contract work under an independent mentor
Built and delivered web and AI work on real client projects, mentored by an experienced developer who owned the client relationships.
- Shipped responsive full-stack web apps with React, Next.js, Node, Express and MongoDB against real client requirements and deadlines.
- Built AI automation with Python and LLM APIs — chatbots and agents that take actions in a workflow rather than only answering questions.
- Handled the parts that decide whether software survives contact with users: REST API design, auth and role-based access, third-party integrations, SEO and performance work.
- Deployed backend services on AWS and connected them to client systems through APIs and webhooks.
- Worked to review: my mentor set the requirements, checked the code, and held the delivery bar — which is where most of what I know about writing maintainable code came from.
- React
- Next.js
- Node.js
- MongoDB
- Python
- LLM APIs
- AWS
Graduated 2026
Lahore
BS Software Engineering
University of Central Punjab (UCP)
Graduated with a 3.61 / 4.00 CGPA. Coursework in data structures, algorithms, system design, databases and software engineering process.

03 — work

Selected work.

Three projects written up properly: the constraint, the architecture, the trade-offs I chose on purpose, and what it cost. Everything else is listed underneath.

2025 — present · actively maintained

Live in production

Mohasib

Statutory e-invoicing platform for Pakistani businesses and tax consultants

A production SaaS platform that reduces Pakistan's mandatory FBR digital-invoicing obligation to a two-click workflow. Invoices are transmitted to PRAL in real time and return a government-issued 22-digit invoice number and QR payload; bulk Excel and CSV imports are validated in full before a single row reaches the regulator.

Live: in production, pre-revenue
Bulk: spreadsheet import replaces sequential entry
Multi-client: one consultant login across many businesses
Sandbox: rehearsal path before any real filing

Next.js
TypeScript
Node.js
MongoDB
PRAL / FBR API
Multi-tenant
PWA
i18n

Read case study Live site Source

2025 — 2026

Live in production

ChefOS

Restaurant operating system — QR ordering, kitchen display, inventory and payments

An all-in-one platform that replaces the paper ticket. Diners order from a scan-to-order QR menu with interactive 3D dish previews; orders reach a Kitchen Display System over WebSockets; inventory decrements as dishes are fired; and payments settle through Stripe or Safepay without staff leaving the till.

Live: deployed and operating at chefos.pro
Zero: paper tickets between counter and pass
QR: ordering on the diner's own device — no hardware
Dual: domestic and international payment rails

React 19
Node.js
Socket.io
MongoDB
Redis
Stripe
React Three Fiber

Read case study Live site Source

2026

Deployed demo

IntelliBid

Auction platform with a conversational bidding agent

A live auction marketplace where the notable component is the agent. Users describe an intent in natural language — a budget, a category, a ceiling — and BidMind, an assistant built on Google Gemini, searches, filters and places incremental bids on their behalf, explaining each action it takes. Bidding itself runs over WebSockets with interactive 3D previews of the lots.

Agent: bids autonomously within a user-set ceiling
Explained: every automated action carries its rationale
Live: WebSocket bidding with server-ordered writes
3D: interactive lot inspection before bidding

React
Node.js
WebSockets
LLM agent
3D previews
MongoDB

Read case study Live site Source

2025 — 2026

Shipped

StudentSphere

Campus platform unifying a marketplace, study materials and a social feed

A single portal replacing the scattered group chats, notice boards and resale posts a university actually runs on. It combines a peer-to-peer marketplace settled through Stripe, shared study materials and courses, and a social feed — all behind one identity and one moderation surface.

3-in-1: marketplace, materials and feed behind one login
Stripe: settled peer-to-peer transactions, no card data held
Unified: moderation across every content type
MERN: end-to-end ownership of the full stack

React
Express
Node.js
MongoDB
Socket.io
Stripe

Read case study Source

Also built

Containerised CI/CD Pipeline
Multi-environment delivery pipeline with Git Flow and staged deployment
- GitHub Actions
- Docker
- Nginx
Source
Mission Control
Local-first personal engineering dashboard
- HTML
- JavaScript
- Local-first
Source
Algorithm Pattern Library
Worked solutions organised by pattern rather than by problem
- C++
- Algorithms
- Data structures
Source

04 — approach

How I work.

Positions I hold about building software, each one traceable to a decision in the work above.

Correctness lives on the server
Hidden buttons are a courtesy; authorisation is a server concern. Every interface I have shipped re-checks permissions where the query is built, because a boundary enforced in the client is not a boundary.
Name the trade-off, then take it
Every architectural choice costs something. I would rather write down what a decision buys and what it forfeits than pretend a design has no downside — the case studies on this site are organised around exactly that.
Design for the failure, not the demo
The interesting work in a regulated or money-handling system is what happens when the other side declines. Recoverable states, persisted exchanges and explicit status beat an optimistic path that only reads well in a screenshot.
Build second, buy first
Payments, identity and compliance primitives are solved problems with real consequences for getting them wrong. The strongest decision in two of these projects was choosing not to build something.
Write it down
If a decision cannot be explained in a paragraph, it is not yet understood. Documenting the reasoning has caught more of my own mistakes than any tool I have used.