# Malik Mohammad Ammar — Full Stack AI Engineer Portfolio

> I build at the intersection of LLMs, computer vision, and robotics.

## About

My work spans agentic AI applications, real-time vision systems, and robotic platforms. I've built AI-powered search engines using reasoning agents and retrieval-augmented generation, developed recommendation systems with machine learning and NLP, deployed object detection and tracking models on NVIDIA Jetson devices, and worked with ROS 2, PX4, and embedded hardware to bridge software and the physical world.

Beyond AI, I enjoy building complete products-from backend infrastructure and databases to modern web applications. Whether it's an AI search engine, a computer vision system, or a consumer-facing platform, I'm interested in turning complex ideas into practical tools that people can use.

I'm particularly interested in how systems can reason, perceive, and interact with the world-bringing together advances in language models, computer vision, and robotics to solve real problems.

## Skills

### Programming Languages
Python, C/C++, JavaScript, TypeScript, Verilog

### Machine Learning / AI
PyTorch, TensorFlow, Scikit-learn, NumPy, Pandas, OpenCV, Hugging Face, LangChain

### Web Development
React, Next.js, Node.js, Tailwind CSS, shadcn/ui

### Mobile Development
React Native, Expo

### Backend & Cloud Platforms
Supabase, Render, Streamlit

### Robotics & Simulation
ROS 2, PX4, AirSim

### Edge AI / Deployment
NVIDIA Jetson, TensorRT, DeepStream

### Developer Tools
Git, GitHub, Docker

## Experience

### Research Engineer — School of Interdisciplinary Engineering & Sciences (SINES) (May 2025 – Aug 2025)
Developed and deployed computer vision algorithms on the NVIDIA Jetson platform for object detection and ID-based object tracking with Re-ID, enabling autonomous behaviour. Built a low-latency video pipeline via GStreamer and OpenCV.

### Intern — Public Sector Research Lab (Jul 2023 – Sep 2023)
Built a hybrid skin-care recommendation system combining user-feature filtering and content-based similarity using TF-IDF and cosine similarity. Achieved 98.38% classification accuracy with Precision@10 = 1.00 and NDCG@10 = 1.00.

## Education

### National University of Sciences and Technology (NUST) — Bachelor of Engineering — Electrical Engineering (2020 - 2024)

## Projects

### OmniFinder AI
Agentic AI search engine with ReAct reasoning and multi-source RAG. Routes queries across Wikipedia, ArXiv & the web using intelligent classification — all through a Streamlit interface.
- GitHub: https://github.com/ammarmalik17/omnifinder-ai

### Salert
Pakistan's deal discovery platform aggregating sales, promotions, and discounts on clothing, food, and consumer items. Features geo-targeted deals, brand following, and voting system.
- GitHub: https://github.com/ammarmalik17/Salert

### Autonomous Quadrotor Object Tracking System
Deployed object detection and Re-ID tracking on NVIDIA Jetson using TensorRT, ONNX, and GStreamer/OpenCV. Validated autonomous quadrotor control via SITL and HITL simulations.
- GitHub: https://github.com/ammarmalik17

### Smart Cane for Visually Impaired
Developed on Raspberry Pi 5 using Python and OpenCV for real-time object detection. Integrated ultrasonic sensors for obstacle detection and audio alerts for safe navigation.
- GitHub: https://github.com/ammarmalik17

### Product Recommendation System
Built a hybrid recommendation system using NLP-based user profiling and content-based similarity with TF-IDF and cosine similarity.
- GitHub: https://github.com/ammarmalik17/skin-care-recommendation-system

### Heart Disease Severity Classification
Performed EDA using box plots, pair plots, and correlation matrices. Identified Triglycerides Level as key predictive feature via PCA, achieving 94.4% accuracy with an SVM model.
- GitHub: https://github.com/ammarmalik17

## Social

- GitHub: https://github.com/ammarmalik17
- LinkedIn: https://linkedin.com/in/malikmammar
- Email: malikmammar17@gmail.com