- Problem
- Setting up GoHighLevel for each new real estate client meant a day of manual edits across workflows, pages, and message templates.
- What I built
- A client-agnostic GoHighLevel snapshot that deploys a full lead-to-close system for real estate clients in under an hour.
- Stack
- GoHighLevel
- Conversation AI
- HTML email
- A2P 10DLC
- Result
- Per-client onboarding went from a day of manual edits to under an hour. Every asset rebrands from one table of 30 Custom Values.
Ali Raza, Full-Stack AI Engineer
I turn AI prototypes into complete, usable software.
Full-stack apps, voice agents, RAG systems, and automation pipelines on n8n, Make.com, and GoHighLevel.
- A caller speaks in Urdu, English, or SpanishLiveKit Agents + Silero VAD
- The agent detects the language and replies in itGroq Whisper + ElevenLabs
- It collects guest name, room type, and phone numberLLM function calling
- The reservation is loggedn8n + Google Sheets
- A WhatsApp confirmation goes out instantlyEvolution API
- to onboard a real estate client onto the GoHighLevel snapshot, down from a day of manual edits
- Under 1 hour
- plus 53 custom fields and 32 SMS/email templates in one reusable snapshot
- 14 workflows
- speed-to-lead response built into the CRM automation
- Under 60 seconds
- handled by the hotel voice agent: Urdu, English, and Spanish
- 3 languages
- end-to-end ML/DL projects published
- 10+
- Next.js
- React
- TypeScript
- Python
- FastAPI
- PostgreSQL
- Redis
- Docker
- LangChain
- n8n
- Make.com
- Flutter
- GoHighLevel
- LangGraph
- LiveKit Agents
- Groq
Featured case studies
Six systems, each one built end to end. Every case study covers the problem, what I built, the stack, and the result.
- Problem
- A hotel has to answer booking calls in the caller's own language, then get each reservation recorded and confirmed.
- What I built
- A real-time voice receptionist that books rooms in Urdu, English, or Spanish and confirms them on WhatsApp.
- Stack
- Python
- LiveKit Agents
- Groq
- ElevenLabs
- Silero VAD
- n8n
- Google Sheets
- Evolution API
- Result
- A caller speaks Urdu, English, or Spanish, books a room by voice, and gets an instant WhatsApp confirmation. Every reservation is logged to Google Sheets, and the agent recovers gracefully if the booking webhook fails.
- Problem
- Teams need one place for workspaces, projects, tasks, and files, where access depends on each person's role and updates arrive in real time.
- What I built
- A full-stack collaboration platform for workspaces, teams, projects, tasks, files, and analytics, with Groq-powered insights.
- Stack
- Python
- FastAPI
- React
- PostgreSQL
- Redis
- Celery
- Docker
- Result
- A containerized platform you can open as a live demo: secure sign-in, role-aware access, real-time updates, and Groq-powered insights, with background reports and CSV exports.
- Problem
- A chat demo only feels like a product once replies stream, conversations persist, and the user can stop or regenerate a response.
- What I built
- A production-style chat assistant with token-by-token streaming and persistent conversations.
- Stack
- React
- TypeScript
- FastAPI
- LangGraph
- PostgreSQL
- Docker
- Result
- A live assistant on Groq-hosted gpt-oss-120b that streams token by token, keeps conversations, and supports stop/regenerate and Markdown.
- Problem
- Scheduling, email, documents, expense tracking, and social publishing are repetitive tasks spread across separate tools.
- What I built
- An n8n personal assistant agent and a content distribution engine that publishes to YouTube and Instagram Reels.
- Stack
- n8n
- Make.com
- Groq
- LangChain
- Google Workspace APIs
- Meta API
- Result
- One agent handles Google Calendar, Gmail, Google Docs, and Sheets expense tracking, and a Google Sheets-driven engine auto-publishes media to YouTube and Instagram Reels.
- Problem
- Questions about official university documents and long YouTube videos need answers grounded in the source, not guesses.
- What I built
- RAG assistants that answer from official IUB PDFs and YouTube transcripts.
- Stack
- LangChain
- PGVector
- PostgreSQL
- Hugging Face
- Groq
- Streamlit
- Result
- Two assistants that answer from official IUB PDFs and YouTube transcripts, using PGVector, MMR retrieval, and Llama 3.1 via Groq behind Streamlit interfaces.
More projects
RazaMind AI
The assistant embedded in this portfolio. It answers visitor questions about these projects.
Next.js, Tailwind CSS
Multi-tool ReAct agent
A conversational agent that picks the right tool for each query: web search, live weather, currency conversion, or a Wikipedia summary.
LangGraph, Groq, Streamlit
Reusable FastAPI authentication service
An async REST API boilerplate with JWT and OAuth2 authentication, bcrypt password hashing, and Aerich database migrations.
FastAPI, Tortoise ORM, Aerich, JWT/OAuth2, SQLite
10+ end-to-end ML/DL projects, published with code.
Scikit-learn, TensorFlow, PyTorch
Technical skills
- Full-Stack & Platform Engineering
- Python, FastAPI, React, Next.js, TypeScript, JavaScript, Vite, Tailwind CSS, Flutter, Dart, REST APIs, WebSockets, JWT/OAuth2, Role-aware authorization, Pydantic
- Generative AI & LLMs
- RAG pipelines, Self-RAG, CRAG, Agentic systems, Prompt engineering, Function calling, Embeddings, Semantic search, MMR retrieval, LangChain, LangGraph, LangSmith, ReAct, Groq API, Anthropic and OpenAI models, Hugging Face, Llama models
- Voice AI
- LiveKit Agents, Silero VAD, Whisper speech-to-text (Groq), ElevenLabs multilingual TTS, Real-time conversational agents, Multilingual language mirroring
- Data, Persistence & Background Processing
- PostgreSQL, PGVector, Redis, Celery, Tortoise ORM, Aerich, SQLAlchemy, SQLite
- Machine Learning, NLP & Vision
- Scikit-learn, TensorFlow, PyTorch, Pandas, NumPy, TF-IDF, Sentiment analysis, Classification, Clustering, SVM, Random Forest, Decision Trees, K-Means, Naive Bayes, Regression, CNNs, LSTMs, Transfer learning, ResNet50
- Workflow Automation & Deployment
- n8n, Make.com, GoHighLevel (CRM architecture, funnels, workflows, Conversation AI, A2P 10DLC compliance), Evolution API (WhatsApp), Docker Compose, Vercel, Streamlit, Git, GitHub, Jupyter Notebook, Google Colab, Kaggle, OpenAPI/Swagger
Experience
Machine Learning Engineer Intern
Developers Hub Corporation (Remote), 2025
Built an end-to-end IMDB sentiment system using TF-IDF features, linear classifiers, and a Streamlit interface; developed ML workflows for NLP, classification, and clustering with reproducible prediction pipelines.
Machine Learning Engineer Intern
Code Alpha (Remote), 2025
Delivered projects in heart-disease prediction, agricultural image recognition, and speech-emotion recognition, training logistic-regression, CNN, and LSTM models with structured preprocessing and performance analysis.
Education
BS Software Engineering
The Islamia University of Bahawalpur, 2023–2027
FSc Pre-Engineering
KIPS College, 2021–2023
Certifications and professional development
Automation & CRM
Delivered a production-ready GoHighLevel (GHL) snapshot system covering CRM structure, funnels, workflows, Conversation AI, and compliance; completed n8n and Make.com training for agentic AI and business automation pipelines.
Generative AI & LLM Engineering
Completed CampusX coursework in LangChain, RAG pipelines, and ReAct agents.
ML/DL & Backend Engineering
Completed Hope to Skill and Siddhardhan ML/DL tracks, plus FastAPI, Tortoise ORM, JWT, PostgreSQL, Redis, Celery, and Docker training.
One Million Prompters
Dubai Future Foundation and Dubai Centre for Artificial Intelligence.
Book a call
A 30-minute intro call. Tell me about the role or the project, your stack, and your timeline, and I will tell you how I would build it.
- aliraza.ai.dev3@gmail.com
- +92 324 5892281
- Multan, Pakistan