Skip to content
Portrait of Muhammad Ibrahim Afzal

Islamabad, Pakistan · GlobeWyze Inc.

Ibrahim Afzal

Software Engineer

Lead Software Engineer shipping regulated healthcare SaaS, fintech platforms, and AI automation. NestJS, Next.js, AWS Bedrock, Stripe, GCP Cloud Run.

ResumeGitHubLinkedInContact
Scroll to explore

About

Engineering production systems end to end

Full stack, backend, AI integrations, and mobile — with ownership from architecture to launch.

I build production systems where backend rigor, product clarity, and modern UX meet. At GlobeWyze Inc., I lead engineering on a Canadian PHIPA-oriented healthcare SaaS—architecting NestJS APIs, Prisma data models, Supabase multi-role auth, Stripe Connect payments, Socket.IO real-time queues, and Next.js portals on GCP Cloud Run.

Alongside regulated healthcare delivery, I ship fintech experiences with AWS Bedrock RAG assistants, AI accounting automation on Lambda and S3, and multi-LLM productivity platforms with Jira and Google Workspace integrations. I authored launch readiness and soft-launch audit documentation used for executive release decisions.

My foundation spans mobile engineering—Flutter, Firebase, and on-device TensorFlow Lite for AR Magic Lens (Final Year Project, Open House selection)—and personal projects spanning FastAPI model serving, computer vision, and Electron desktop apps.

I combine backend depth, mobile experience, and product-minded execution—shipping systems that hold up under real users, real payments, and real compliance constraints.

3.62

CGPA

400+

Git Commits

5

Coursera Certs

Healthcare SaaS

Technical Lead

Education

B.S. Software Engineering, NUML Islamabad

Jun 2025 · CGPA 3.62/4.00

FYP: AR Magic Lens — Open House selection

Experience

Shipping regulated, payment-heavy products

Technical leadership and full-stack delivery across healthcare, fintech, AI automation, and mobile.

NestJSPrismaSupabaseNext.js 14StripeSocket.IOGCP Cloud Run
  • Technical lead for Canadian PHIPA-oriented healthcare SaaS: architecture direction, MVP scope, backend + frontend coordination, and production deployment readiness.
  • #1 engineering contributor — 73 backend + 91 frontend commits (164 total); 30 Prisma models and 27+ migrations with Controller → Service → DB service layering.
  • Built Supabase Auth BFF, multi-role RBAC, Stripe Connect payment lifecycle, Socket.IO provider queues, QR intake, PDF verification, and admin/clinic portals.
  • Owned GCP Cloud Run CI/CD (Bitbucket Pipelines, OIDC, Docker) and authored launch readiness and soft-launch verification audit documentation.
Next.js 16SupabaseOpenAIAnthropicBedrockJiraGoogle Calendar
  • 38 commits as primary builder: Supabase auth, RLAs, floating time-log UI, pause/resume timers, and tenant isolation fixes.
  • Built multi-provider AI service (OpenAI, DeepSeek, Anthropic, OpenRouter, Bedrock) with JSON extraction from LLM responses.
  • Integrated Jira and Google Calendar/Workspace one-way sync with background job queue and chat history persistence.
Next.js 15NestJSAWS BedrockS3SQSApp RunnerRedux
  • Built Fynn Agent from scratch and integrated it into the private-markets platform.
  • 226 commits on frontend: IC Agent FAB, AI Assistant UI, Bedrock RAG via RTK Query, S3 presigned uploads, and investment-community workflows.
  • 13 commits on backend AI service: Amazon Bedrock Agent Runtime, RAG service, IC-Agent APIs, RBAC, and cross-user data leakage remediation.
  • Deployed microservices on AWS App Runner via ECR with GitHub Actions; Route 53 DNS configuration for production endpoints.
PythonFastAPILambdaNext.js 15Claude VisionPrismaSAMpgvectorDeepSeekOpenAI
  • Primary author (32/36 commits, 89%) of Python pipeline: dual OCR (Tesseract + Claude Vision), transaction extraction, Canadian GIFI tax-code Excel reports.
  • Primary author (23/24 commits, 96%) of Next.js 15 frontend: 5-step workflow, Supabase auth, categorization rules engine with learning API, Google Sheets export.
  • Deployed accounting automation via AWS Lambda, S3, SAM/CloudFormation, and Amplify; ~10,758 lines added across the engagement.
  • Built FastAPI agent with pgvector RAG on Supabase Postgres, DeepSeek LLM, and OpenAI embeddings for deal workflows.
  • Intent router (create_deal / ask_question / general_chat) with confirm-gated deal-draft schemas for Real Estate, VC, and Private Credit.
  • Next.js 15 developer console for chat testing, citations, and admin knowledge-base ingest (PDF/DOCX/TXT/MD).
FlutterDartFirebaseTensorFlow Lite
  • Developed AR Magic Lens — Flutter mobile app for object recognition and pronunciation for children.
  • Delivered production MVP in 8 weeks using on-device TensorFlow Lite (not ARCore/ARKit).
  • Final Year Project selected for university Open House.

Skills

Depth across the full stack

From regulated backend systems to AI integrations and mobile engineering.

Languages

TypeScriptJavaScriptPythonDartSQLC++Java

Projects

Production systems with measurable ownership

Healthcare SaaS, fintech AI, accounting automation, and mobile AR — each with real engineering depth.

Canadian Healthcare SaaS Platform

Regulated medical absence notes · Technical Lead

164 commits30 models27+ migrations

Challenge

Deliver a PHIPA-oriented healthcare platform with multi-role portals, real-time provider queues, payment flows, and production launch readiness under regulatory constraints.

Solution

Led full-stack architecture with NestJS layered APIs, Supabase BFF auth, Stripe Connect lifecycle, Socket.IO queues, QR intake, PDF verification, and Next.js role-based portals on GCP Cloud Run.

Architecture

Controller → Service → DB service layering across 30 Prisma models; Supabase JWKS JWT with multi-role RBAC; Bitbucket Pipelines CI/CD with OIDC to Cloud Run.

Impact

164 commits as #1 contributor; authored launch readiness and soft-launch audit documentation for executive go/no-go decisions.

NestJSPrismaSupabaseNext.js 14Stripe ConnectSocket.IOGCP Cloud Run

Private-Markets / Investment Platform

Fintech · AWS Bedrock RAG

226 FE commitsBedrock RAG

Challenge

Integrate AI assistants into investment-community workflows while securing multi-tenant data and deploying microservices at scale.

Solution

Built Bedrock RAG experiences via RTK Query, S3 presigned document uploads, SQS messaging, and NestJS AI microservices with RBAC and leakage remediation.

Architecture

Next.js 15 frontend + NestJS ai-service on AWS App Runner via ECR; Bedrock Agent Runtime with OpenSearch Serverless for RAG retrieval.

Impact

226 frontend commits delivering IC Agent FAB, AI Assistant UI, and production-hardened tenant isolation fixes.

Next.js 15NestJSAWS BedrockS3SQSApp RunnerRedux

AI Accounting Document Automation

Finance document pipeline · Primary Author

89% BE authorship96% FE authorship

Challenge

Transform bank and credit-card PDFs/images into structured, tax-compliant Excel reports with high OCR accuracy and batch throughput.

Solution

Dual OCR pipeline (Tesseract + OpenCV + Claude Vision) with quality-based selection; Claude transaction extraction; Canadian GIFI mapping; Next.js 15 BFF with rules-learning API.

Architecture

Python CLI → FastAPI on AWS Lambda + S3 (SAM/CloudFormation); Next.js 15 API routes as BFF to FastAPI; Prisma 7 + PostgreSQL with Supabase auth.

Impact

Primary author: 89% backend (32/36) and 96% frontend (23/24) commits; ~10,758 lines added Oct–Dec 2025.

PythonFastAPILambdaClaude VisionNext.js 15PrismaSAM

RLA & Time-Tracking SaaS

Conversational AI productivity platform

38 commits5 LLM providers

Challenge

Unify time tracking, compliance proof packages, and multi-provider AI assistance with enterprise integrations and strict tenant isolation.

Solution

Built Supabase auth, RLAs, floating time-log UI, multi-LLM abstraction (OpenAI, DeepSeek, Anthropic, Bedrock), and Jira + Google Calendar sync.

Architecture

Next.js 16 + Supabase + Zustand; background job queue for sync; JSON extraction layer across LLM providers with chat history persistence.

Impact

38 commits as primary builder covering onboarding, approvals, analytics (CSV/PDF), and invoice flows.

Next.js 16SupabaseOpenAIAnthropicBedrockJira API

FastAPI Investment RAG Agent

Standalone AI agent prototype

Sole author

Challenge

Prototype an intent-aware investment assistant with grounded retrieval, deal drafting, and a developer console for rapid iteration.

Solution

FastAPI agent with pgvector on Supabase Postgres, DeepSeek LLM, OpenAI embeddings, and confirm-gated deal-draft actions across asset classes.

Architecture

Intent router (create_deal / ask_question / general_chat) + Next.js 15 dev console for citations and KB ingest (PDF/DOCX/TXT/MD).

Impact

Sole author (4 commits) delivering end-to-end RAG pipeline and testing console.

FastAPIpgvectorDeepSeekOpenAINext.js 15

AR Magic Lens

FYP + Internship · Flutter AR

8-week MVPOpen House FYP

Challenge

Help children learn object recognition and pronunciation through an accessible mobile AR experience without platform-native AR SDKs.

Solution

Flutter app with on-device TensorFlow Lite object detection, Firebase backend, and text-to-speech pronunciation feedback.

Architecture

Camera pipeline + TFLite inference + Firebase auth/storage; delivered as 8-week internship MVP and Open House FYP.

Impact

Shipped production MVP in 8 weeks; selected for university Open House presentation.

FlutterDartFirebaseTensorFlow LiteTTS

Additional projects

Emotion Detection (FER2013)

Personal project · Computer vision

Challenge

Serve a trained emotion recognition model with low-latency inference from a modern web interface.

Solution

FastAPI + PyTorch model serving with Next.js webcam UI; containerized deployment on Render with Docker.

Architecture

Separate inference API and React frontend; Docker Compose for local dev; production Docker image on Render.

Impact

End-to-end ML serving pipeline from training artifact to browser inference.

FastAPIPyTorchNext.jsDockerRender

EyeAura

CCTV vision prototype

Challenge

Prototype a vision system for face detection and recognition in a controlled CCTV workflow.

Solution

Next.js control plane with FastAPI vision worker using InsightFace, ONNX Runtime, and OpenCV; Supabase for data layer.

Architecture

Decoupled Next.js dashboard + Python vision microservice with ONNX model inference pipeline.

Impact

Demonstrated full-stack CV system design from UI to inference worker.

Next.jsFastAPIInsightFaceONNXOpenCVSupabase

Tech Stack

Tools I ship with in production

A constellation of technologies across cloud, backend, frontend, and AI.

TypeScriptNext.jsNestJSPythonPostgreSQLPrismaSupabaseAWS BedrockLambdaS3GCP Cloud RunDockerStripeSocket.IOFlutterTensorFlow LiteOpenAIClaudeTailwind CSSFastAPITypeScriptNext.jsNestJSPythonPostgreSQLPrismaSupabaseAWS BedrockLambdaS3GCP Cloud RunDockerStripeSocket.IOFlutterTensorFlow LiteOpenAIClaudeTailwind CSSFastAPI

Frontend

  • TypeScript
  • Next.js
  • React
  • Tailwind CSS

Backend

  • NestJS
  • Node.js
  • FastAPI
  • Python
  • Socket.IO

Data

  • PostgreSQL
  • Prisma
  • Supabase
  • Firebase

Cloud & AI

  • AWS Bedrock
  • Lambda
  • S3
  • GCP Cloud Run
  • Docker
  • OpenAI
  • Claude

Mobile

  • Flutter
  • Dart
  • TensorFlow Lite

Certifications

Continuous learning

Coursera credentials across AI, Python, frontend, UX, and project management.

AI For Everyone

DeepLearning.AI

Coursera

Programming for Everyone (Python)

University of Michigan

Coursera

Introduction to Front-End Development

Meta

Coursera

Foundations of UX Design

Google

Coursera

Project Management Essentials

Howard University

Coursera

Contact

Let's build something production-ready

Open to full stack, backend, mobile, and AI integration roles.

Reach out for opportunities, collaborations, or technical conversations.

ibrahimafzal00@hotmail.com
+92-343-5173768LinkedInGitHub