Fakhar-e-Iqbal Abdullah

AI Architect | Agentic Systems & RAG Specialist | Building Production-Grade AI Solutions

Based in: PakistanCurrently: AI Agents Architect, Independent Product DevelopmentExperience: 4 years

AI Architect and Engineer with a proven track record of deploying production-grade agentic systems and RAG pipelines. From building autonomous SOC analyst platforms to scaling Saudi legal assistants with 90% accuracy, I specialize in bridging the gap between cutting-edge research and containerized, cloud-native deployments. I thrive on solving complex data integration challenges and building AI that delivers measurable business impact.

Experience

AI Agents Architect

Independent Product DevelopmentfounderMay 2026 – Present

Summary

Designing an AI business advisor that gives honest, fact-based advice, using real data on competitors, market trends and businesses that failed or succeeded. Building a system that joins a company's own data (product, marketing, social media, team) with live research from outside the business. Splitting the analysis across several cooperating AI agents, each with its own job, so that together they produce one clear recommendation.

What I did

Developed an orchestration layer where a primary agent collects SMB details and connects to Shopify inventory via Admin API or direct database access.
Built specialized agents for business ads account analysis, social media post monitoring, and Slack team communication tracking.
Integrated data collection workflows to feed internal metrics into a research agent layer for competitor analysis.
Implemented schema validation by wrapping model calls in a retry loop to check data matching and re-prompting the model upon failure.

AI AgentsMarket AnalysisData IntegrationShopify Admin APISlack APIInstructor

Freelance AI Engineer, Software Engineer & SaaS Full-Stack Developer

Freelancefreelance2023 – Present

Summary

Delivered AI, machine learning and full-stack projects for international and local clients, from requirements to deployment, and kept long-term client relationships across freelance platforms and direct contracts. Built SaaS products and web and mobile apps end to end (React, NestJS, Flutter) with OpenAI, Gemini and Claude APIs. Traffic Sign Detection: Trained YOLO models to detect and read traffic signs for an autonomous driving team (about 92% mAP), sending the results as signals into their system. Dental X-ray Reports: Built a classifier (about 94% accuracy) that passes its findings to an LLM to write a patient report. Location Research Tool: Built a tool where you pick an area and a business type, and it finds competitors within 2 to 3 km, collects Google and web reviews, and ranks them. Created AI visuals for a Unity game and an AI content-to-video pipeline for a kids' cartoon YouTube channel. Gold Mines Machine Age Prediction for Botswana Client: Combined sensor, log and temperature data from several industrial sources into one dataset, then trained an XGBoost regression model (R² of about 0.88) to estimate machine age. Autonomous Job Application Agent(n8n, Gmail, LinkedIn, Jooble): Built a two-pipeline agent that finds job openings, extracts HR contact details and sends personal applications with Gemini-written cover letters. Reached a contact discovery rate above 90% and handled 50+ applications a day.

What I did

Dental X-ray Reports: Built a classifier that passes its findings to an LLM to write a patient report.
Gold Mines Machine Age Prediction for Botswana Client: Combined sensor, log and temperature data from several industrial sources into one dataset, then trained an XGBoost regression model to estimate machine age.

Results

Traffic Sign Detection: about 92% mAP.
Dental X-ray Reports: about 94% accuracy.
Gold Mines Prediction: R² of about 0.88.
Autonomous Job Agent: Reached a contact discovery rate above 90% and handled 50+ applications a day.

ReactNestJSFlutterOpenAIGeminiClaude APIYOLOXGBoostn8nUnityYOLOv8FastAPI

AI Architect & Engineer

NextGeneration Technology Solutions, LahoreLahoreworkFeb 2026 – May 2026

AI SOC Analyst Platform (Claude Agent SDK, PostgreSQL, Wazuh): Built an autonomous agent pipeline that handles Wazuh security alerts from start to finish. It extracts indicators of compromise, checks them against threat intelligence, categorizes the alert and writes an investigation plan. Cut manual triage time by about 60%. Built a threat intelligence module that pulls live data from 6 feeds (OpenPhish, URLhaus, ThreatFox, OTX, CISA KEV, SANS ISC), using a two-level PostgreSQL design for fast filtering and pagination at scale. Added crash recovery, retry logic and SOP-driven investigation plans matched to alerts through a 16-SOP keyword registry. AI HR System: Designed a complete AI-powered HR system for a US fashion client that handles routine HR tasks end to end. Mentored interns and led the team's engineering work, keeping quality and delivery on track.

Claude Agent SDKPostgreSQLWazuhOpenPhishURLhausThreatFoxOTXCISA KEVSANS ISCPydanticFastAPI

Skills

Technical

Python

LangChain

LlamaIndex

YOLOv5/v8

Hugging Face Transformers

Anthropic Claude Agent SDK

PyTorch

TensorFlow

MLflow

DVC

Faster R-CNN

n8n

Flutter

U-Net

ResNet

LangGraph

PostgreSQL

Docker

AWS

Kubernetes

React

NestJS

GitHub

XGBoost

YOLOv8

Pydantic

FastAPI

Instructor

Projects

Saudi Legal RAG Assistant

AI/ML Engineerprofessional

Built an Arabic/English multimodal RAG agent for a Saudi client that answers questions over 2,000+ pages of Saudi law, accepts images, and supports voice in and out through OpenAI. Used hybrid retrieval (BM25 plus vector search).

OpenAIBM25Vector SearchRAGFastAPI

Autonomous Job Application Agent

Developerpersonal

Built a two-pipeline agent that finds job openings, extracts HR contact details and sends personal applications with Gemini-written cover letters.

n8nGmailLinkedInJoobleGemini

AI-Powered Food Ordering App

AI Engineerprofessional

Built a real-time ordering app with a RAG chatbot that helps users explore menus, place orders and get personal recommendations. Added route-based ordering that times food preparation to the user's arrival.

FlutterGeminiVector Database

Traffic Sign Detection

AI Engineerfreelance

Trained YOLO models to detect and read traffic signs for an autonomous driving team, sending the results as signals into their system.

YOLOYOLOv8

Gold Mines Machine Age Prediction

AI Engineerfreelance

Combined sensor, log and temperature data from several industrial sources into one dataset, then trained an XGBoost regression model to estimate machine age for a Botswana client.

XGBoostPythonSensor Data

AI SOC Analyst Platform

AI Architect & Engineerprofessional

Built an autonomous agent pipeline that handles Wazuh security alerts from start to finish. It extracts indicators of compromise, checks them against threat intelligence, categorizes the alert and writes an investigation plan.

Claude Agent SDKPostgreSQLWazuhPydanticInstructor

Education

Bachelor of Science

COMSATS University IslamabadComputer Science2021 – 2025

Societies & activities

Google Developer Students Club

Recognition

Certifications

Machine Learning Specialization

DeepLearning.AI (Andrew Ng)

Advanced Learning Algorithms

DeepLearning.AI (Andrew Ng)

Unsupervised Learning, Recommenders, Reinforcement Learning

DeepLearning.AI (Andrew Ng)

Supervised Machine Learning: Regression and Classification

DeepLearning.AI (Andrew Ng)

Social impact & activities

AI & Data Science Lead

Google Developer Students Club, COMSATS University Islamabadworkshop

Designed and taught 'AI Engineering & Data Science' to 40+ students, covering machine learning, deep learning and data science, and mentored members on their projects.

Hire Fakhar-e-IqbalReach out about a role, a contract or a conversation.For recruiters

Fakhar-e-Iqbal's twin is AI, it can make mistakes.

More people on Kanvis

Browse more people Create your own Kanvis