Welcome to my GitHub profile! I'm Ammar, a Data Scientist with a passion for crafting intelligent systems, building scalable solutions, and solving complex problems with the power of data and AI. Let me walk you through my journey, the technologies I love working with, and some of the exciting projects I've been fortunate to work on!

• Programming: Python is my go-to language for most tasks, but I also enjoy working with C++, JavaScript, and even Bash scripts when needed.
• Frameworks and Tools: From building web apps with Django and Flask to deploying them with Docker and managing infrastructure with Terraform, I enjoy bringing ideas to life end-to-end.
• AI/ML: I specialize in Computer Vision, NLP, and Deep Learning using tools like TensorFlow, PyTorch, and advanced techniques like Vision Transformers, Stable Diffusion, and Retrieval-Augmented Generation (RAG).
• Cloud Platforms: I leverage AWS, Azure, and Google Cloud to deploy scalable AI solutions.
• Data Pipelines: I thrive on automating data processing workflows with tools like SQL, BigQuery, and ETL pipelines.
• Visualization and Insights: I enjoy transforming raw data into actionable insights using Power BI, Tableau, and Looker Studio.

• I’ve built systems that automate ETL workflows, reducing manual intervention by 95%. Think of scripts that orchestrate AWS Lambda or Google Cloud Functions to handle massive datasets seamlessly.
• In another project, I developed a Recommendation Agent powered by RAG and clustering techniques, boosting engagement rates significantly. I even optimized the algorithms using cuML/RAPIDS, cutting processing time by 40%.

• Using Facebook’s Segment Anything Model 2 (SAM2) and Flask, I created a pipeline to track objects in videos. Imagine breaking down a video into frames with FFmpeg, applying advanced AI models, and deploying it on AWS EC2 to deliver multi-object tracking efficiently.

• Ever wanted an AI to organize your photo gallery? I developed a photo culling algorithm leveraging MobileNetV3 and DBSCAN to detect duplicates, closed eyes, and even prioritize aesthetically pleasing images. It’s a tool that saves countless hours for photographers.

• I worked on forecasting fuel prices using LSTM, XGBoost, and Neural Prophet, achieving impressive accuracy. These time series models processed years of historical data to predict trends for the next few weeks.

• Whether it’s creating a chatbot using a decoder-only transformer model or designing a semantic search system using embeddings, I’ve applied NLP to solve real-world challenges. One of my projects even connected SQL databases with NLP models for intelligent query processing.

• I explored the creative side of AI with Stable Diffusion, deploying models that generated realistic deep fake images and replicated faces with high precision. I also trained LoRA models to add stylistic features, pushing the boundaries of generative AI.

• Imagine a chatbot that understands not only your queries but also retrieves relevant data efficiently. I developed one using Django, LlamaIndex, and Pinecone, combining RAG with traditional SQL databases to create a robust AI assistant.

• Enhancing image quality has been a fascinating challenge. I worked on a project using TensorFlow and Keras to upscale hyperspectral and RGB images by up to 8x, significantly improving clarity.

• I containerized a Flask and React app with Docker, automated deployments using GitHub Actions, and managed infrastructure with Terraform. It was an exciting blend of software engineering and DevOps practices.

There’s something magical about combining creativity and logic to solve problems, whether it’s training an AI model to interpret data or deploying a solution that impacts users directly. I believe in constantly learning and experimenting with new technologies to push the limits of what’s possible.

• GitHub: github.com/101ammarAh
• LinkedIn: Ammar Ahmad

Feel free to explore my projects, collaborate on open-source work, or just say hi. Let’s innovate together! 🚀