Full-Stack Software Engineer & AI Systems Specialist with 4+ years of experience building scalable web applications, distributed microservices, and production-grade AI systems.

• 🔭 Currently: Engineering full-stack AI platforms, multi-agent frameworks, and high-throughput web systems.
• 💼 Experience: Marqeta Inc. (NASDAQ: MQ), BIS Research, Full-Stack & AI Systems Consultant (Contract/Freelance).
• 🛠️ Core Tech: React.js, Next.js, Node.js, Express, Python (Django/Flask), TypeScript, GraphQL, PostgreSQL, AWS, & Docker.
• 🎓 Education: B.Tech in Computer Science & Engineering (YMCA University).
• 📫 How to reach me: shujarahman786@gmail.com
• 🌐 Connect: LinkedIn | GitHub

Languages & Frameworks
Python Django Flask Node.js Express.js Javascript TypeScript React.js Next.js GraphQL Redux Tailwind CSS

AI, LLM Benchmarking & Automation
OpenAI / Groq / Gemini APIs Prompt Engineering AI Agent Orchestration Evaluation Rubrics pytest Bash/Shell Scripting

Cloud, DevOps & Databases
AWS (EC2, S3, Lambda, Amplify, CloudFront) Docker GitHub Actions PostgreSQL MongoDB Redis Elasticsearch

• Freelance AI Evaluation Engineer (Outlier & 3rd-Party Platforms)

  • Designed evaluation benchmarks, complex coding/debugging tasks, and grading rubrics for frontier open-source & proprietary LLMs across single-agent and multi-agent workflows.
  • Authored oracle solutions, test suites, and specification-driven validation pipelines to ensure high benchmark reliability.

• Gatekeeper Rules Engine Benchmark

  • Built a reproducible LLM evaluation benchmark for an authorization engine featuring recursive inheritance, policy precedence, and temporary overrides.
  • Wrapped in isolated Docker execution environments with automated pytest validation pipelines.

Multi-agent AI travel planner built with React 18, Node.js, TypeScript, Turborepo, Prisma, PostgreSQL, & Dual LLM Providers.

• Engineered a sequential multi-agent execution pipeline (Interpreter ➔ Destination ➔ Itinerary ➔ Budget Agents) to translate natural language prompts into structured travel plans.
• Implemented dynamic failover between Google Gemini and Groq APIs for fault tolerance, secured JWT authentication, and managed state using Prisma ORM with Neon PostgreSQL.

Real-time AI-powered author support platform built with React 19, Node.js, TypeScript, PostgreSQL, & AWS.

• Features automated AI classification and priority routing for incoming tickets using Groq LLMs and prompt engineering with context chunking.
• Implemented real-time bi-directional messaging via Socket.IO, TanStack Query server-state management, and deployed to AWS (Amplify, Elastic Beanstalk, CloudFront).

AI-powered document workflow and compliance management platform.
Backend Repo | Frontend Repo

• Automates document processing, AI-driven metadata extraction, approval workflows, compliance tracking, and audit logging.
• Built with secure, role-based document operation architecture and scalable analytics pipelines.

Reltio-inspired Master Data Management suite.

• Features fuzzy matching, entity resolution, golden record creation, and interactive relationship graph visualization using React Flow and Socket.io

Tech Stack: React.js Node.js TypeScript GraphQL Apollo Router Python Groq LLM Salesforce API Datadog SonarQube

• Distributed Systems & Performance: Architected full-stack features across microservices using React.js, Node.js, and GraphQL, cutting page load times by 25%. Optimized cross-service routing with Apollo Router, driving a 30% boost in API throughput.
• Enterprise AI & Support Automation: Engineered an LLM pipeline using Groq AI, context chunking, and advanced prompt engineering to automatically classify support tickets and draft responses, saving 10+ hours/week in operations.
• Process Automation & Integration: Built Python microservices and integrated Salesforce APIs to automate account provisioning and communication, eliminating 20+ hours/week of manual administrative overhead.
• Reliability & Observable Infrastructure: Implemented Datadog monitoring and integrated SonarQube quality gates into CI/CD pipelines, driving down production incidents by 12%.

Tech Stack: Next.js TypeScript Redux Node.js Python Elasticsearch MongoDB GPT-3.5 Stripe Highcharts

• AI Search & Discovery: Built Insight Navigator, combining GPT-3.5 and Elasticsearch to provide contextual Q&A and sub-second querying across multi-million record datasets.
• Performance Optimization: Leveraged Next.js SSR/SSG, dynamic imports, and lazy loading to increase Lighthouse performance by 40%.
• Automated Payments & Scheduling: Streamlined enterprise subscription flows and consulting bookings by integrating Stripe API and Calendly.
• Product Analytics: Created custom behavioral analytics dashboards using Highcharts, Hotjar, and HubSpot to inform strategic product roadmap decisions.