Syed shah abdur Rehman

Full-Stack Software Engineer & AI Agent Architect

Based in: Lahore, PakistanMost recently: Web Developer, Tbox SolutionsExperience: 4 years

A versatile software engineer specializing in building voice-enabled AI agents, generative UI research, and enterprise-grade full-stack platforms using Python, TypeScript, and React. He has a proven track record of delivering complex systems, from conversational booking agents with WebRTC to robust data pipelines and location-aware marketplaces.

Experience

Web Developer

Tbox SolutionsworkFeb 2025 – Aug 2026

Built, debugged and deployed Laravel/JavaScript applications across 5+ client projects, adapting to unfamiliar codebases, integrations and deadlines. Designed MySQL-backed data layers and REST integrations, using Git for collaborative development and review.

LaravelJavaScriptMySQLRESTGit

Software Engineer

NETSOL TechnologiesLahore, PakistanworkSep 2025 – Jun 2026

Summary

Built full-stack features and REST endpoints for enterprise finance platforms using React and FastAPI, working across multiple services within agile sprint cycles. Reviewed and debugged shared code with 10+ engineers, discussing performance, maintainability and reliability; documented decisions for technical and non-technical teammates. Researched Generative UI using Vercel AI SDK and Google A2UI; used Claude Code, Cursor and MCP workflows to support implementation and code review.

What I did

Resolved integration issues between the frontend and backend

Results

Worked on an enterprise finance platform with a team of 10+ engineers

ReactFastAPIRESTVercel AI SDKGoogle A2UIClaude CodeCursorMCP

Tutor and Web Developer

Superprofwork2023 – 2025

Delivered React/Node.js applications with PostgreSQL and MongoDB from requirements through deployment; taught international students to understand unfamiliar code and debug logic errors.

ReactNode.jsPostgreSQLMongoDB

Skills

Technical

Gemini Live

JavaScript

Tool calling

TypeScript

Python

FastAPI

Pipecat

React

Cursor

PostgreSQL

SQLite

MySQL

MongoDB

WebRTC

MCP

Anthropic APIs

OpenAI APIs

Claude Code

Vercel AI SDK

Google A2UI

Next.js

Playwright

React Native

Tailwind CSS

Node.js

OpenClaw

PHP

Django

Laravel

AWS Bedrock

Docker

scikit-learn

pandas

NumPy

AWS EC2

SeleniumBase

xvfb

systemd

Caddy

coturn

Fine-tuning

Areas of expertise

Agile/Scrum

Projects

Generative UI Research

Researcherpersonal

Researched how agents can turn a conversational request into an interactive interface, connecting LLM tool outputs to usable product workflows. Studied Vercel AI SDK approaches to rendering tool results in React and Google A2UI approaches to describing UI through structured, declarative messages. Research focus: component selection, structured output contracts, streaming updates, client-side rendering and routing user actions back to an agent. Considered how these patterns could make complex workflows accessible through conversation while keeping application behavior controlled.

Vercel AI SDKGoogle A2UIReactLLM

Agent Tooling and ASO Automation

Freelance Developerfreelance

Customized an open-source agent framework with new tool integrations and AWS Bedrock model support for a freelance client. Connected browser automation to agent workflows so research could lead to concrete, reviewable output. Built a Playwright-based tool that collected competitor app data from Google Play through an ASO tracking extension. Combined Sensor Tower/MCP tooling, competitor keywords and app metadata to support ASO analysis and listing drafts. Documented setup, tool behavior and workflow steps so the system could be understood, reproduced and maintained beyond the initial delivery.

OpenClawMCPPlaywrightAWS BedrockSensor Tower

WhatsApp Battery Pricing Agent

Developerfreelance

Built a data pipeline for a solar battery company that turns unstructured dealer-group messages into validated product and price records. Applied a cheap prefilter before LLM extraction, with a strict JSON contract for identifying priced battery listings. Deduplicated normalized messages and product records, updated changed prices, and persisted processing state in SQLite across restarts. Wrote daily group-specific Google Sheets output through a swappable LLM provider interface. Added offline chat replay with dry-run and provider modes to inspect extracted records, dropped messages and duplicate handling before live use.

TypeScriptBaileysLLM APIsSQLiteGoogle Sheets

AroundYou Marketplace and Ordering Agent

Developerpersonal

Developed and published a location-aware marketplace to Google Play. Merchants define service areas on a map; customers discover and order from businesses within range, with merchant-specific delivery pricing and driver management. Built a customer-side AI agent for ordering and cart management, connecting conversational assistance to application actions and the shopping workflow.

React NativeAI AgentGoogle Play

Visit the site

UK Company Enrichment Pipeline

Data Engineerpersonal

Designed a resumable pipeline for a 126,858-company source set, combining discovery, official-site verification, bounded crawling and selective LLM review. Stored stage-level evidence and model decisions in SQLite, with checkpoints, retries and a durable writeback queue for recovery after interruption. Separated verified source evidence from model judgments and provided evaluation against gold data, supporting inspectable results rather than untraceable generated records.

PythonSQLiteNocoDBLLMSeleniumBaseAWS EC2xvfbResidential IPs

NIMBESS Data Platform

Backend Developerprofessional

Built backend integrations for battery telemetry and product workflows. Verified signed vendor events against the exact request bytes and committed raw notifications before acknowledgment, protecting data from a provider with no retry or historical replay API. Used durable nonce constraints, role-scoped APIs, approvals and audit events to connect reliable ingestion with device and operational workflows.

FastAPIPostgreSQL

Construction ERP and BlackStone CRM

Full Stack Developerprofessional

Designed and delivered multi-role business systems with access controls and audit logs. Implemented project-scoped inventory, invoicing, payroll and financial dashboards in the ERP; customer/vendor invoice and credit views, inventory and salary workflows in the CRM. Owned requirements, data modeling, implementation and deployment, translating operational needs into interfaces and permission rules that clients could use.

ReactNode.jsMongoDBREST APIs

Driver Drowsiness System

personal

Built a system using YOLO for eye detection and a blink rate counter to monitor driver alertness.

YOLO

Vibecoderzz Voice Booking Agent

Lead Developerpersonal

Built and deployed a voice assistant on my portfolio that answers service questions, qualifies enquiries and books consultations. Connected a React/TypeScript interface to a FastAPI/Pipecat service over WebRTC, with Gemini Live handling the conversation and server-side tools performing calendar actions. Integrated Cal.com availability and booking tools with time-zone and business-hour validation. Required explicit visitor confirmation; reported a booking as successful only after the API returned a booking identifier. Stored transcripts, lead details and booking progress for an admin dashboard. Added bounded response deadlines, reconnect handling, interrupted-reply tracking and cleanup so stalled calls recover or expose an actionable error. Verified the deployed flow in Chromium under packet loss, offline recovery and relay-only transport. Tested reply continuity at 15% packet loss and documented the remaining TURN/TLS-on-443 infrastructure requirement.

PythonPipecatGemini LiveNext.jsReactTypeScriptFastAPIWebRTCCal.comChromiumVercelAWS EC2systemdCaddycoturn

Visit the site

Crypto Sentiment Analysis Model

personal

Developed a sentiment analysis model using GPT-2 to analyze tweets for foundational crypto trading analysis.

GPT-2

Education

BS Computer Science

Bahria University, LahoreComputer Science2022 – 2025

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

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

More people on Kanvis

Browse more people Create your own Kanvis