Hi, I'm Mohammad Ibrahim
AI/ML Engineer
Passionate about developing responsible and impactful AI solutions. I specialize in machine learning, large language models, and computer vision to create intelligent systems that solve real-world problems.
Passionate about developing responsible and impactful AI solutions. I specialize in machine learning, large language models, and computer vision to create intelligent systems that solve real-world problems.
Get to know me better
I am an aspiring AI/ML Engineer with over 10 months of hands-on experience in developing and deploying intelligent systems. I hold a B.Tech in Computer Science and Engineering and am currently working as an AI/ML Engineer at PharynxAI, where I contribute to real-world applications involving Gen AI and computer vision.
Passionate about responsible and impactful AI, I also serve as a Global Ambassador for the Responsible AI UAE Chapter, where I actively promote ethical practices in AI development and deployment. My areas of interest include Artificial Intelligence, Machine Learning, Large Language Models (LLMs), chatbot development, fine-tuning open-source LLMs, OpenCV, and object detection.
Projects Completed
AI/ML Domains
Technologies Used
Developing and deploying ML models for various applications using TensorFlow and other frameworks.
Building intelligent systems including chatbots, AI agents, and generative AI applications.
Creating object detection and image processing solutions using OpenCV and deep learning.
Fine-tuning large language models and developing natural language processing applications.
My professional journey
Representing the UAE in global AI initiatives and contributing to international AI policy discussions and collaborations.
Actively promoting ethical practices in AI development and deployment, organizing events and workshops on responsible AI.
Contributing to real-world applications involving Gen AI and computer vision, developing and deploying intelligent systems for various use cases.
Exploring the intersection of AI, machine learning, and real-world applications
Designed and developed the backend for an AI-powered SEO tool that enables marketing teams to optimize and rank their products across multiple large language models (LLMs), streamlining workflows and improving product visibility.
Designed and developed an AI-powered email management assistant that enables professionals to automatically categorize, summarize, and schedule emails across Google services, streamlining inbox management and improving productivity through intelligent calendar integration.
Designed and developed an AI-powered cosmetic safety tool that scans product labels using OCR, extracts ingredients, and cross-references them with FDA (US) and Indian regulatory databases to identify banned or harmful substances, ensuring transparency and consumer safety.
End-to-end Python pipeline for fine-tuning LLMs on custom content. It parses PDFs to Markdown via Docling, chunks them by headings, generates Q&A pairs using vLLM (Qwen3-8B), and uses these pairs for fine-tuning. Includes automated inference for evaluating model performance post-training.
Trained a Rasa NLU model for intent classification and entity extraction using custom training data. Supports multiple user intents such as greetings, FAQs, and service inquiries with high accuracy.
RESTful API using FastAPI to generate detailed job descriptions and KPIs from role summaries, integrated with LLM backend (Ollama API with LLaMA 3.1 model) for AI-powered natural language generation.
Complete AI system using spaCy, SentenceTransformers, and Ollama for parsing and matching resumes with job descriptions using semantic embeddings and similarity algorithms.
Built a machine learning pipeline to predict number plate prices in the UAE by scraping data from Xplate.com. Included analysis of patterns like repeating digits, culturally significant numbers (e.g., 786), and short sequences.
Advanced LSTM neural networks trained on Nifty-50 stock data with technical analysis, moving averages visualization, and deployed as interactive Streamlit web app.
Fine-tuned Seq2Seq model using FLAN-T5-large with Hugging Face transformers for advanced abstractive text summarization of legal documents.
Tools and technologies I work with
Professional credentials and achievements
Let's work together on your next project