Shafiq ur rehman

Shafiq ur rehman

AI Software Engineer specializing in Backend, Cloud Infrastructure, and Agentic AI Systems.

Based in: Lahore, PakistanCurrently: Independent AI & Cloud Engineering Projects, Self-DirectedExperience: 1 year

Hire me

AI Software Engineer with over 2 years of experience building scalable backend services and deploying containerized applications via AWS and GitHub Actions. Expert in full-cycle development, from API design and microservices architecture to advanced LLM training and RAG implementation.

Experience

Independent AI & Cloud Engineering Projects

Self-DirectedLahore, PakistanfreelanceMar 2026 – Present

Summary

Designed and deployed containerized ML and AI services end to end on AWS EC2 with GitHub Actions CI/CD, automated pytest suites, and MLflow tracking, alongside advanced certifications in microservices, Kubernetes, RAG, and agentic AI.

What I did

  • Separated the model from the API using FastAPI to make the ML pipeline easier to test, reproduce, and deploy.

Results

  • Achieved 0.985 recall and 0.9996 ROC-AUC for a model deployed as a containerized ML service on AWS EC2.
  • Integrated automated testing into the CI/CD pipeline using GitHub Actions and pytest.
AWS EC2GitHub ActionsCI/CDpytestMLflowMicroservicesKubernetesRAGAgentic AI

DSA Trainer & Mentor

Dev Weekends Fellowship ProgramRemote, Pakistanpart timeMay 2025 – Present
Sustained a 15-month mentoring engagement, delivering Data Structures & Algorithms (C++) training to 300+ students with one-on-one mentoring and progress assessments.
C++Data StructuresAlgorithms

AI Engineer & LLM Trainer

TuringRemote, United StatescontractOct 2025 – Mar 2026

Summary

Authored, solved, and evaluated 150+ competitive programming problems in Python and C++ over 4 months, averaging 2 accepted problems per day for RLHF-based model training. Validated AI-generated code against official documentation and isolated edge cases and logical gaps, supplying corrective solutions that improved code-generation reliability.

Results

  • Averaged 2 accepted problems per day for RLHF-based model training.
  • Improved code-generation reliability through corrective solutions.
PythonC++RLHFLLM Training

Skills

Technical

Data Structures & Algorithms
REST APIs
JavaScriptJavaScript
GitHub ActionsGitHub Actions
CI/CD
PythonPython
C++C++
AWS (EC2, S3, IAM)
OOP
Node.jsNode.js
DockerDocker
Express.jsExpress.js
OpenAI API
LinuxLinux
pytest
Git/GitLabGit/GitLab
FastAPIFastAPI
Microservices
RedisRedis
JWT / OAuth 2.0
TypeScriptTypeScript
SQLSQL
System Design
MongoDBMongoDB
LLM Training & Evaluation
RLHF
RAG
LangChain / LangGraph
MLflow
React.jsReact.js
Next.jsNext.js
React NativeReact Native
Redux Toolkit / RTK QueryRedux Toolkit / RTK Query
Tailwind CSSTailwind CSS
MySQLMySQL
Event-Driven Architecture
FirebaseFirebase
MCP Agents
KubernetesKubernetes
VercelVercel
QLoRA Fine-Tuning
Distributed Systems
pandas
XGBoost
PostgreSQLPostgreSQL
NginxNginx
Vue.jsVue.js
SQLiteSQLite
Reflexion
Role-Based Access Control (RBAC)

Projects

Multi-Vendor E-Commerce Platform

Full Stack Developerpersonal
Full-stack marketplace with self-service vendor registration, multi-category catalog and order management, and checkout supporting card and cash-on-delivery payments; containerized and deployed on AWS EC2.
Next.jsNode.jsExpressMongoDBReduxDockerAWS EC2JWTRole-Based Access Control (RBAC)

FraudGuard - End-to-End ML Fraud Detection System

Lead Developerpersonal
Built and deployed a real-time transaction fraud scoring API trained on millions of transactions, reaching 0.985 recall and 0.9996 ROC-AUC. Owned the full pipeline: MLflow tracking, pytest suite, Docker containerization, GitHub Actions CI/CD, and AWS EC2 deployment behind a Vercel-hosted client.
FastAPIXGBoostDockerGitHub ActionsAWS EC2MLflowpytestVercel

E-Learning LMS Platform

Full Stack Developerpersonal
Industry-level LMS with JWT authentication, DRM-protected lecture streaming via VdoCipher, and separate student and admin dashboards; deployed on AWS EC2, with Redis caching cutting repeat API response time by 40%.
Next.jsTypeScriptNode.jsExpressMongoDBRedisRTK QueryJWTVdoCipherAWS EC2

Reflexion-based AI Agent

personal
Developed an agent that uses a multi-step self-correcting loop where it generates an answer, reviews its own response for mistakes, and regenerates based on that feedback.
Reflexion

Education

Bachelor of Science in Computer Science

University of EducationComputer ScienceAug 2022 – May 2026

Coursework

Data Structures & AlgorithmsOperating SystemsComputer NetworksDatabase SystemsSoftware EngineeringArtificial Intelligence

Recognition

Awards

Bronze Fellowship

Dev Weekends2025

Honorable Mention, ICPC Asia Topi Preliminary Contest

ICPC2024

Competitive programming contest achievement

LeetCode 100 Days Badge

LeetCode

250+ problems solved

Certifications

Microservices with Node.js and React

Udemy2026

AI Engineer Core Track (LLM Engineering, RAG, QLoRA Fine-Tuning)

Udemy2026

Agentic Track (Agents & Model Context Protocol)

Udemy2026

Patents & publications

Retrieval-Augmented Generation techniques and limitations

publication

Technical article on Dev.to

Read it

Reflexion-based AI agents with LangGraph

publication

Technical article on Dev.to

Read it

Model selection for production workloads

publication

Technical article on Dev.to

Read it

React performance internals

publication

Technical article on Dev.to

Read it

Social impact & activities

Performance and Scalability Sessions

Dev Weekendsworkshop

Delivered 10+ recorded sessions covering caching, load balancing, horizontal scaling, and fault tolerance.

Authentication and Authorization Sessions

Dev Weekendsworkshop

Delivered sessions on JWT and OAuth 2.0.

DSA Problem-Solving Sessions

Dev Weekendsworkshop

Delivered sessions on Data Structures and Algorithms problem-solving.

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

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

More people on Kanvis

Browse more peopleCreate your own Kanvis