Hi, I'm Mohammad Afrid Pasha

Problem Statement

Traditional fitness assessments require expensive equipment, trained personnel, and physical contact, preventing scalable remote athlete evaluation

Solution Approach

Architected production-grade AI platform using MediaPipe pose estimation, MiDaS depth networks, and PyTorch models achieving 95% accuracy with contactless measurements (±2cm height, ±3kg weight)

Key Features

• Real-time pose estimation at 30 FPS with sub-100ms latency
• Contactless body measurements with 95% accuracy
• Exercise classification with 92% precision across 4+ exercise types
• ML-powered athlete benchmarking and clustering
• Live analytics dashboard with performance visualizations
• Automated leaderboard and progress tracking

Problem Statement

Urban traffic congestion causing delays and inefficient traffic flow management

Solution Approach

Implemented YOLO v8 computer vision for vehicle detection with ML optimization

Key Features

• Real-time vehicle detection
• Traffic flow optimization
• Smart signal control
• Analytics dashboard

Problem Statement

Lack of platform connecting local service providers with consumers efficiently

Solution Approach

Developed MERN stack platform with microservices architecture and hybrid database

Key Features

• Dual marketplace
• Service provider matching
• Real-time booking
• Payment integration

Problem Statement

Enterprise file transfer systems are fragmented, protocol-locked, and fail to deliver simultaneous high-throughput transfers, real-time P2P collaboration, and zero-knowledge security within a single scalable platform

Solution Approach

Engineered a production-grade, cloud-agnostic distributed platform unifying HTTP REST, WebRTC DataChannels, raw TCP Sockets, and WebSocket streams — backed by CRDT-powered collaborative IDE, AES-256-GCM zero-knowledge encryption, multi-cloud storage abstraction (Cloudflare R2, Backblaze B2, MinIO), and Redis Pub/Sub achieving 800 Mbps throughput with sub-25ms P99 latency

Key Features

• Quad-protocol architecture — HTTP, WebSocket, WebRTC P2P DataChannels & raw TCP streams in a single platform
• Real-time collaborative CodeShare IDE powered by CRDT differential sync with live multi-cursor tracking
• AES-256-GCM zero-knowledge client-side encryption with Argon2id hashing and TOTP 2FA (Google Authenticator)
• Multi-cloud VFS abstraction with Circuit Breaker failover across Cloudflare R2, Backblaze B2, and MinIO
• Redis-backed Token Bucket rate limiting, Bloom Filter O(1) lookups, and HyperLogLog cardinality estimation
• Merkle Tree chunk integrity verification with resumable uploads and SHA-256 end-to-end checksums
• Secure OTP share links with Fibonacci backoff brute-force mitigation and UUIDv4 IDOR protection
• Kubernetes HPA auto-scaling with Prometheus/Grafana observability and structured JSON audit logging

Problem Statement

Luxury perfume retailers lack intelligent customer support and secure payment systems for online shopping experiences

Solution Approach

Architected scalable MERN stack platform integrating Google Gemini AI chatbot, Stripe payments, and microservices architecture with auto-scaling on Render Cloud

Key Features

• AI chatbot with voice recognition reducing support queries by 40%
• Secure Stripe payment gateway with PCI-DSS compliance
• Real-time inventory management and order tracking
• Promo code system with automated discount validation
• Responsive UI with 3D animations and <2s load time
• Role-based access control with JWT authentication

Problem Statement

Organizations facing increasing cyber threats with slow detection and response times

Solution Approach

Engineered ML-powered threat detection system processing 1M+ events per second

Key Features

• Real-time threat detection
• Automated response system
• Network monitoring
• Security analytics