AI Ticketing System
RAG · Enterprise Platform
AI-powered ticket management with context-aware support recommendations
Spring BootSpring SecuritySpring AIReactTypeScriptViteMySQLQdrantGroq (LLaMA 3.3 70B)JWTSwagger/OpenAPI
- Developed a full-stack AI-powered ticket management system combining enterprise ticketing with Retrieval-Augmented Generation (RAG) for context-aware support recommendations.
- Built a secure REST API using Spring Boot, Spring Security, and JWT, with a React + TypeScript + Vite frontend.
- Integrated Spring AI with Groq Llama 3.3 70B and Qdrant Vector Database to retrieve semantically similar historical tickets and generate AI-powered summaries, category predictions, priority classification, and suggested resolutions.
- Designed a layered Spring Boot architecture with DTOs, mappers, service, repository, and specification patterns.
- Implemented secure JWT authentication and role-based authorization using Spring Security.
- Built a Retrieval-Augmented Generation (RAG) pipeline using Qdrant for semantic similarity search over historical support tickets.
- Developed RESTful APIs with pagination, sorting, dynamic filtering, validation, and global exception handling.
- Documented APIs using Swagger/OpenAPI.
Proposal AI Generator
Multi-Agent System
Slack-native proposal automation · built in one week
LangGraphGroq (LLaMA 3.3 70B)QdrantSlack SDKFastAPISQLitepython-docxStreamlit
- Built a three-agent pipeline that converts client discovery call transcripts into fully formatted DOCX business proposals, delivered directly in Slack within seconds of upload.
- Orchestrated stateful agent handoffs with LangGraph: one agent extracts structured client data, one retrieves relevant context via RAG, and one synthesizes the final proposal.
- Implemented semantic retrieval over Qdrant with industry metadata filtering, so retail clients surface retail-relevant past proposals.
- Designed composite session keys (user_id:channel_id) for concurrent multi-client proposal management with cross-session memory.
- Added targeted section-level editing — follow-up requests update only the relevant section and redeliver the revised DOCX automatically.
SYNC — Influencer Marketing Platform
Final Year Project
FAST-NUCES Karachi · Spring 2026
Django RESTFastAPIReact 19TypeScriptGroq LLMMySQLMeta APIJWTCelery
- Architected a full-stack three-tier platform (React SPA + Django REST API + FastAPI ML microservice) enabling AI-assisted discovery and collaboration between businesses and influencers.
- Built an ML matching engine using Groq API (Llama 3.2) for zero-shot niche and popularity vector prediction; two-stage ranking with cosine similarity (50%) and budget-price overlap (50%).
- Engineered a complete collaboration lifecycle: negotiation, escrow-style payment hold, dual-side completion marking, and site-admin payout release with automated 10% fee calculation.
- Implemented credit-gated identity reveal with field-level encryption and JWT-based role separation.
- Integrated Meta Instagram API (OAuth) for live profile ingestion with hybrid pricing model.
Patient Registration Voice AI System
Voice AI · Full-Stack Platform
Voice-driven patient registration & records management, backed by a live dashboard
VapiFastAPIDjango RESTMySQLReactTypeScriptVitePydantic
- Built an end-to-end patient registration platform where a Vapi voice AI agent holds natural conversations with patients to register, look up, and update their records.
- Architected the pipeline as Patient → Vapi Voice AI → FastAPI → Django REST API → MySQL → React Dashboard, cleanly separating voice orchestration, business logic, and data storage.
- Used FastAPI as an AI middleware layer to execute tool calls from the voice agent and used Pydantic models to validate and structure data extracted from free-form conversation.
- Built the backend APIs and patient data storage with Django REST Framework and MySQL, exposed to a React + TypeScript + Vite management dashboard for staff.
- Try the live voice registration workflow by calling +1 (346) 209-5384 (limited demo call credits available).
Autism Predictor Bot
Computer Vision
EfficientNetV2-SPyTorchTransfer LearningGroqGradio
- Developed a deep-learning image classifier using EfficientNetV2-S transfer learning for early autism screening, with an LLM explanation layer for contextual prediction insights.
- Built end-to-end pipeline covering preprocessing, training, evaluation, and Gradio deployment.
Plumbing Guide RAG Chatbot
RAG System
LangChainChromaDBRAGVector Embeddings
- Built a retrieval-augmented chatbot with structured document chunking, conversational memory, and optimized vector-embedding retrieval for multi-turn troubleshooting support.
Assignment Feedback Agent
Agentic AI
LangChainRAGAgentic AIMCP OrchestrationTool Calling
- Built an agentic AI system evaluating student assignments via rubric-based scoring and structured LLM reasoning pipelines with tool-calling capabilities.
- Designed an MCP-style orchestration flow: agent decomposes tasks, selects tools, evaluates outputs, and iteratively refines feedback.
