Shafiq Ur Rehman
AI Software Engineer | Cloud, DevOps & Backend
I build full-stack and agentic AI products, and I ship them to production. AWS, Docker, CI/CD, microservices.
I build products end to end and put them live. I've shipped a learning platform with locked-down paid lectures and a marketplace where sellers run their own stores. Most recently I deployed and operated FraudGuard, a fraud scoring service that runs on AWS and redeploys itself every time I push code. I also build AI agents with LangChain, LangGraph, and MCP, and I spent six months at Turing on the problems behind code-reasoning models.
About
What I build, and where I'm heading next.
My Journey
I'm a full-stack and AI engineer working across cloud and DevOps, based in Lahore. I started with web apps in Next.js, Node.js, and MongoDB, and the projects I'm proudest of are the ones people actually used.
At some point the question I cared about changed. It stopped being "does this feature work" and became "what happens when a hundred people use it at once, and what happens when it breaks at 3am." That pulled me toward infrastructure. Now I containerize what I build, automate the testing and deploys, and run it on AWS. Kubernetes is next, so I can run services across a cluster instead of one machine.
The AI work sits alongside it, and I'm more interested in the systems around the models than in the models themselves. At Turing I spent six months writing and reviewing hard coding problems used to train models to reason better about code. I build agents with LangChain, LangGraph, and MCP, write about them on Dev.to, and teach algorithms to 300+ students at Dev Weekends.
What I bring
Ships to production
Containers, automated tests, CI/CD pipelines, and a live AWS deployment on every project I finish.
Backend and APIs
REST APIs in Node.js, Express, and FastAPI, with Redis caching and JWT or OAuth 2.0 auth.
Agentic AI
RAG pipelines, LangChain and LangGraph agents, and MCP, plus six months of RLHF data work at Turing.
Teaches and writes
Fifteen months mentoring 300+ students in DSA, with 10+ recorded sessions and six published articles.
Where I'm heading
Moving my deployed services off single-instance Docker and onto orchestrated Kubernetes clusters, and contributing upstream in the CNCF ecosystem while working toward GSoC and LFX Mentorship.
Technical Skills
AWS
Docker
Kubernetes
GitHub Actions
CI/CD
Nginx
Node.js
Express
Redis
MongoDB
MySQL
Firebase
Next.js
ReactReact Native
Vue.js
TypeScript
Python
C++
JavaScript
LangChain
OpenAI APIProfessional Journey
Where I've worked and what I shipped there.
Independent AI & Cloud Engineering
Self-Directed
Building and deploying containerized AI services end to end.
Key Achievements
- Own the full pipeline: FastAPI services, Docker images, pytest suites, GitHub Actions CI/CD, and AWS EC2 deployment
- Use MLflow for run and artifact tracking across deployments
- Completed certifications in microservices with Kubernetes, RAG, and agentic AI (MCP), then applied each one to shipped work
Technologies Used
AI Engineer & LLM Trainer
Turing (Contract)
Authored, solved, and evaluated 150+ competitive programming problems in Python and C++ over six months, averaging two accepted problems a day, as training and evaluation data for RLHF. Validated model generated code against official documentation and found edge cases and logical gaps where the model's reasoning broke down.
Key Achievements
- Contributed to Large Language Model (LLM) training using RLHF (Reinforcement Learning with Human Feedback) by creating, solving, and evaluating competitive programming (CP) problems.
- Solved and developed algorithmic and problem-solving tasks in Python and C++, focusing on correctness, efficiency, and edge-case handling.
- Provided high-quality solutions, feedback, and corrections to train and evaluate AI models, improving reasoning and code generation quality.
Technologies Used
Software Engineer Intern
Stem Solutions
Shipped 12 production screens for a cross-platform Android and iOS app in React Native and TypeScript. Built a reusable component library and cut initial screen load time by 25% through render optimization.
Key Achievements
- Developed frontend interfaces for web and mobile applications during a structured internship program, focusing on responsive design and user experience.
- Collaborated with development teams using Git version control and agile methodologies
- Implemented reusable components and optimized frontend performance.
Technologies Used
DSA Trainer & Mentor
Dev Weekends Fellowship Program
Fifteen months running DSA (C++) training for 300+ students: weekly sessions, one-on-one mentoring, progress tracking. Published 10+ recorded sessions covering scalability fundamentals, authentication and authorization, React internals, and algorithm patterns.
Key Achievements
- Delivered structured DSA training sessions to 300+ students.
- Provided one-on-one mentorship to improve coding techniques and problem-solving skills.
- Conducted progress assessments and provided constructive feedback.
Technologies Used
Featured Projects
Backend services, infrastructure, and the applications I've shipped.
FraudGuard
Deploying and operating a fraud scoring service
I took a fraud detection model and built everything around it that makes it a real running system: a containerized FastAPI endpoint, pytest covering the request paths, and GitHub Actions deploying to AWS EC2 on every push.
E-Commerce Platform
A marketplace where vendors self-register and manage their own catalogs. Multi-category product support, order management, and checkout with both card payments and cash on delivery. Containerized and deployed on AWS EC2.
Aurora ChatBot
A MERN commerce app with a built-in conversational assistant that answers product questions and surfaces catalog items, plus an admin panel for managing products, categories, and customer inquiries in one place.
Elearning
A learning platform with JWT auth and DRM protected lecture streaming through VdoCipher, plus separate student and admin dashboards. Redis caching cut repeat API response time by 40%. Deployed on AWS EC2.
Horizon Impact Fund Managers
A paid client build for a Southern Africa impact investment firm, presenting the business, focus areas, and ESG story. JWT auth with 2FA and role-based access guards internal areas, backed by Redis and RTK Query.
Expense Tracker
A React Native expense tracker for Android and iOS with local SQLite storage, category breakdowns, and spending charts that update as you log transactions. Built with TypeScript and a reusable component set.
Writing & Speaking
Long-form technical writing and recorded sessions.
Everyone Is Building AI Agents. Almost Nobody Is Building AI Systems.
What separates a working agent demo from a system that survives real users. Covers the five pillars a production AI system needs beyond the agent loop, state, observability, evaluation, recovery, and security, plus a maturity model and real incidents where each one was missing.
Authentication in MERN Apps: JWT, bcrypt, Redis, and OAuth2
Building authentication that holds up in production. Access and refresh token split, why localStorage is unsafe for tokens, bcrypt cost tradeoffs, Redis for revocation and rate limiting, and Google OAuth2, with breach case studies and a security checklist.
Retrieval-Augmented Generation (RAG): How It Works, 10 Advanced Techniques, and Common Limitations
A practical walkthrough of retrieval-augmented generation: how the retrieval and generation halves fit together, and ten techniques for improving results once a basic implementation stops being good enough. Covers chunking strategy, reranking, and query rewriting, along with the failure modes people hit in production and why more context often makes answers worse rather than better.
From Simple LLMs to Reliable AI Systems: Building Reflexion, Based Agents with LangGraph
The article covers why single-pass LLMs are unreliable and how Reflexion fixes this by making AI agents critique and retry their own outputs. LangGraph provides the infrastructure to wire that self-correction loop together.
How to Choose the Right AI Model for the Right Job
Explores different AI models, their strengths and trade-offs, and provides practical guidance on selecting the right model based on the problem, performance, cost, and real-world use case.
How React Achieves High Performance, Even With Extra Layers
Explains how React optimizes performance despite abstraction layers.
JavaScript Execution Context Made Simple
Simplified breakdown of execution context and call stack.
Performance & Scalability Fundamentals
Delivered an in-depth session on performance and scalability, covering latency, throughput, caching, load balancing, horizontal and vertical scaling, fault tolerance, and system design concepts through real-world architecture diagrams.
State Management in React
A deep-dive session on React state management, covering component vs application-level state, prop drilling, Context API, and performance optimizations like Virtual DOM diffing and Fiber's priority-based reconciliation.
Authentication Authorization & Security
Delivered an in-depth session on authentication and authorization, covering session-based authentication, JWT, OAuth 2.0, Google Authentication, security best practices, and interview-focused questions with real-world implementation concepts.
JavaScript Under the Hood
Delivered a live technical session on JavaScript Under the Hood, explaining the JavaScript engine, execution context, call stack, event loop, memory management, and asynchronous programming concepts.
DSA Session: String Theory & Core Concepts
Delivered a session on string concepts, covering common string operations, optimization techniques, and the foundations required for coding interviews.
DSA Problem Solving: Strings with Leetcode Problems
Solved string problems with coding examples, explaining optimized approaches, interview-focused solutions, problem-solving mindset, edge cases, and time and space complexity analysis.
DSA Session: Slow & Fast Pointers Fundamentals
Delivered a session on the Slow & Fast Pointer technique, covering its core concepts, pointer movement strategies, common use cases, and the intuition behind solving coding interview problems efficiently.
DSA Session: Slow & Fast Pointers Problem Solving
Solved Slow & Fast Pointer coding problems step by step, focusing on interview-ready solutions, problem-solving mindset, optimized approaches, and edge cases.
DSA Live Session: Arrays with Leetcode Problems
Delivered a live problem-solving session on LeetCode arrays, solving multiple array problems step by step while explaining the thought process, optimal approaches, and time and space complexity analysis.
Arrays (Part 2) with Leetcode Problems
Conducted a continuation session on array problem solving, covering more LeetCode challenges with a focus on pattern recognition, algorithmic thinking, optimized solutions, and coding best practices.
Certifications & Achievements
Validating technical expertise through recognized certifications and competitive programming success.
ICPC Asia Topi Online Preliminary Contest
ICPC Foundation
Honorable Mention in the 2024 ICPC Asia Topi Online Preliminary Programming Contest.
Dev Weekends Bronze Fellowship
Dev Weekends
Awarded the Bronze Fellowship by Dev Weekends in recognition of steady effort, perseverance, continuous learning, and positive contribution throughout the fellowship journey.
AI Engineer Core Track: LLM Engineering, RAG, QLoRA & Agents
Udemy
Completed an advanced AI Engineering program covering Large Language Models (LLMs), Retrieval-Augmented Generation (RAG), QLoRA fine-tuning, AI agents, vector databases, prompt engineering, and production-ready AI application development.
Microservices with Node.js and React
Udemy
Completed a comprehensive course on building scalable microservices using Node.js and React, covering event-driven architecture, Docker, Kubernetes, authentication, testing, CI/CD, and production-ready distributed systems.
AI Engineer Agentic Track: The Complete Agent & MCP Course
Udemy
Completed an advanced course on building autonomous AI agents using the Model Context Protocol (MCP), agent workflows, tool integration, memory, planning, and multi-agent systems for real-world AI applications.
Get in touch
I'm looking for cloud, DevOps, and backend engineering roles. Open to remote work and to relocation.