I help companies build production-grade AI systems. With 9+ years of experience in AI/ML, I specialize in Document AI, OCR pipelines, Retrieval-Augmented Generation (RAG), and LLM-based agents used in real businesses across banking, healthcare, manufacturing, and e-commerce. If you’re struggling with unstructured data, PDFs, scanned documents, hallucinating LLMs, or brittle prototypes, I can help you take it to production. What I Build: 🔹 Document AI & OCR Systems • Invoice, contract, COA, compliance & regulatory documents • Custom OCR pipelines (AWS Textract, Azure Document Intelligence, PaddleOCR) • Layout-aware extraction using LayoutLM, custom CV + NLP hybrids • Entity extraction, tables, multi-page documents, noisy scans • Deployed at scale in banking & manufacturing environments 🔹 Retrieval-Augmented Generation (RAG) Systems • Enterprise RAG using Azure AI Search, Pinecone, Neo4j • Hybrid search (vector + keyword), metadata filtering, reranking • Query rewriting, HyDE, self-reflection loops to reduce hallucinations • Multi-source retrieval (databases, PDFs, Shopify, Gorgias, internal docs) 🔹 LLM Pipelines & Agentic Systems • Multi-step LLM workflows with evaluators & guardrails • LangChain, LangGraph, Semantic Kernel • Structured outputs with function calling & schema enforcement • Fine-tuned LLaMA-3 (LoRA) for domain-specific extraction & reasoning • Cost-optimized pipelines (caching, async execution, retries) 🔹 Computer Vision • HP Singapore: printer component inspection systems • Object detection, image classification, defect detection • Models deployed in real manufacturing workflows 🔹 Speech AI • Whisper fine-tuning for noisy & impaired speech • Real-time transcription pipelines • Voice cloning & synthesis (ElevenLabs-based systems) Tech Stack: Python, PyTorch, FastAPI, Docker LangChain, LangGraph, Semantic Kernel OpenAI, Claude, LLaMA-3, Whisper Azure AI Search, Azure OpenAI, Azure Document Intelligence AWS Textract, ECS, Lambda Pinecone, Neo4j, PostgreSQL Industries I’ve Worked In Banking • Healthcare • Manufacturing • E-commerce • Food-Tech Why Clients Hire Me • I’ve taken AI systems to production, not just notebooks • I understand failure modes: hallucinations, OCR noise, bad retrieval • I design systems that are scalable, debuggable, and cost-aware • I can work independently and think like an AI architect, not just a coder If you need someone who can design, build, and own an end-to-end AI system, let’s chat.

Want to deploy production-ready AI models, automate complex visual/text workflows, or build rigorous data pipelines? I bridge the gap between advanced research and scalable engineering. As an ML Engineer and Data Scientist with a Master’s degree, I specialize in building deterministic, high-performance AI systems. From fine-tuning LLMs and training computer vision models to engineering leakage-aware research datasets, I deliver clean, reproducible, and production-grade code. Whether you are a startup needing a custom RAG pipeline or an enterprise optimizing massive datasets, I ensure your models perform with mathematical precision. 🚀 CORE CAPABILITIES • AI Systems, LLMs & Automation Building robust RAG pipelines, structured data extraction systems, and automated workflows using Gemini, OpenAI, and Claude APIs. Expert in prompting strategies that yield deterministic, enterprise-grade outputs. • Natural Language Processing (NLP) & Dataset Engineering Experienced with Hugging Face Transformers for text classification, summarization, and multi-label categorization. Specialized in building leak-aware, balanced training/validation pipelines (e.g., CVEfixes vulnerability dataset parsing). • Computer Vision & Synthetic Data Generation Hands-on experience with object detection, image segmentation (U-Net variants), and custom preprocessing. I design synthetic image/video generation pipelines using state-of-the-art tools (Runway, Vidu, Veo, Stable Diffusion) for specialized model training and simulation datasets. • Machine Learning & Deep Learning (PyTorch & TensorFlow) End-to-end model development, validation, and optimization using PyTorch, TensorFlow, and Scikit-Learn. Expert at resolving class imbalances, handling massive datasets, and model benchmarking. • Statistical Modeling & Time Series Deep mathematical background in optimization, logistic/linear regression, hypothesis testing, and time-series forecasting. Experienced with complex real-world data analysis (e.g., NHANES health prediction models). • Data Quality & Engineering Advanced data cleaning and curation using Python, R, CleanLab, and Argilla to isolate and eliminate noise, outliers, and duplicates before they hit your training loop. 🛠️ TECHNICAL TOOLKIT • Languages: Python, R, Bash, SQL • AI/ML Frameworks: PyTorch, TensorFlow, Scikit-Learn, Hugging Face • Data Tools: CleanLab, Argilla, Pandas, NumPy, Excel (Pivot Tables) • Visualization: R (ggplot2), Python (Matplotlib, Seaborn) • Generative Media: Veo, Runway, Vidu, Stable Diffusion API Let’s turn your data into a competitive advantage. Click the "Invite to Job" button to discuss your project requirements!

Top Rated Plus developer with a 100% Job Success Score specializing in high-performance Android (Kotlin/Jetpack Compose) and iOS (Swift/SwiftUI) applications. I provide a competitive edge by integrating on-device AI - from custom fine-tuned YOLO models for real-time detection to local LLM integration for intelligent UX. Simultaneously, I offer expert-level React Native development for clients seeking high-intensity, scalable cross-platform solutions. I bridge the gap between complex AI logic and polished mobile UX. A recent production success includes a YOLO-based scanning app achieving a 100% detection rate and 99% OCR accuracy via TensorFlow Lite. Technical Expertise: - On-Device AI & Vision: Expert in custom YOLO, OpenCV, and Google ML Kit for real-time scanning and on-device LLM features. - Native Excellence: Senior-level SwiftUI and Jetpack Compose development focusing on clean MVVM/MVI architecture. - High-Intensity React Native: Building robust cross-platform apps with seamless native-module integrations and high-performance UI. - Backend & Cloud: Scalable infrastructure using AWS Lambda, Firebase, Supabase, GraphQL, and REST APIs. - Growth & Revenue: Integration of MMPs (AppsFlyer, Adjust) and analytics (Mixpanel, Customerio) to track revenue and user behavior. - Modernization: Expert at migrating legacy codebases to modern frameworks and optimizing app performance. I lead projects from the first line of code to App Store/Google Play deployment, ensuring every app I ship is production-ready, performant, and data-informed.

I build production-ready AI systems that solve real business problems. With 6+ years of experience in AI/ML and Python backend engineering, I help startups and companies design, build, and deploy LLM applications, RAG pipelines, agentic workflows, NLP systems, computer vision solutions, and predictive models. My work covers Python, FastAPI, open-source LLMs, OpenAI integrations, vector search, model serving, evaluation, and cost/latency optimization. Core expertise: LLMs, RAG, Agentic AI, Python, FastAPI, NLP, Computer Vision, model deployment, API integrations, and cloud-based AI systems. If you need someone who can turn an AI idea into a reliable production system, let’s talk. Leul A

Objectives Being an enthusiastic engineer and active researcher in the field of Machine Learning and Robotics Control, I want to develop new AI-driven applications and robotic frameworks that enable humans to lead a better life style. Therefore, I am interested to become a part of an organization or a research group where I can exercise and enhance my skills that help me to achieve my research goals. I am always motivated by the challenges and try to develop something useful for the community. I believe, work until your signature becomes an autograph.

Freelancer - Verified Skills from Upwork Test System I am Muhammad Usman Ghani Khan, a seasoned Data Scientist and AI expert with over 20 𝘆𝗲𝗮𝗿𝘀 𝗼𝗳 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵 𝗮𝗻𝗱 𝗱𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁 experience in machine learning, deep learning, image processing, computer vision, natural language processing, bioinformatics, medical imaging and computational linguistics. 𝗘𝘅𝗽𝗲𝗿𝘁𝗶𝘀𝗲: • 𝗔𝗿𝘁𝗶𝗳𝗶𝗰𝗶𝗮𝗹 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 𝗮𝗻𝗱 𝗠𝗮𝗰𝗵𝗶𝗻𝗲 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴: I earned my PhD in Computer Science (Machine Learning/Artificial Intelligence) from The University of Sheffield, UK. With over 20 years of experience in AI, ML, and data science, I have published more than 200 impact factor publications across various AI domains, including natural language processing, computer vision, and medical applications. • 𝗖𝗼𝗺𝗽𝘂𝘁𝗲𝗿 𝗩𝗶𝘀𝗶𝗼𝗻: I am an expert in modern computer vision techniques. I have led numerous R&D projects in image processing, computer vision, and video analysis. My work includes developing machine learning models for object detection, segmentation, and image classification using 2D and 3D images. • Generative AI: I am a well versed trainer and developer working in the new paradigm of GenAI. Following are some of my expertise in the area of GenAi. Natural Language Processing including (i) text processing (ii) feature extraction (iii) word embedding and (iv) RNNs/LSTMs/GRUs Running LLM'S off the Shelf: LLM APIs, Open-source LLMs, Prompt engineering, Structuring outputs Working with vector databases like Chroma, Pinecone, Milvus etc Retrieval Augmented Generation using LangChain and LlamaIndex Prompt Engineering: Few Shot Prompting, Chain-of-thoughts Prompting Supervised finetuning via LoRA and QLoRA Quantization and Model Distillation Agentic Development: LangGraph and Crew AI etc. Deployment of LLMs: LM Studio, LangServe, AWS EC2 Instance etc. Monitoring of LLMs: LangSmith • 𝗡𝗮𝘁𝘂𝗿𝗮𝗹 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗣𝗿𝗼𝗰𝗲𝘀𝘀𝗶𝗻𝗴 (𝗡𝗟𝗣): I have a deep understanding of NLP techniques and large language models like ChatGPT. My experience includes developing custom NLP solutions for text generation, classification, and information extraction. I have successfully applied NLP to various domains, including biomedical text processing and computational linguistics. • 𝗠𝗲𝗱𝗶𝗰𝗮𝗹 𝗔𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀: With extensive experience in bioinformatics and medical imaging, I develop AI solutions for healthcare, including the analysis of medical images (X-ray, CT, MRI) and biomedical data. My work spans from data preprocessing and feature engineering to model development and deployment. • 𝗗𝗲𝗲𝗽 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗙𝗿𝗮𝗺𝗲𝘄𝗼𝗿𝗸𝘀: Proficient in TensorFlow and PyTorch, I have hands-on experience in building and training custom neural networks for various applications. My expertise includes using these frameworks for computer vision, NLP, and other machine learning tasks. • 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵 𝗮𝗻𝗱 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁 𝗟𝗲𝗮𝗱𝗲𝗿𝘀𝗵𝗶𝗽: As the Director of National Centre of Artificial Intelligence and founder of multiple research labs, including the Computer Vision & Machine Learning Lab, Bioinformatics Lab, Virtual Reality & Gaming Lab, Data Science Lab, and Software Systems Research Lab, I lead cutting-edge research and innovative projects. • 𝗧𝗲𝗰𝗵𝗻𝗶𝗰𝗮𝗹 𝗣𝗿𝗼𝗳𝗶𝗰𝗶𝗲𝗻𝗰𝘆: Experienced in Python, Java, JSON, Django, Asp.Net, R, MATLAB, and Ubuntu. Skilled in using GitHub for version control and LaTeX for documentation. 𝗘𝗱𝘂𝗰𝗮𝘁𝗶𝗼𝗻𝗮𝗹 𝗖𝗼𝗻𝘁𝗿𝗶𝗯𝘂𝘁𝗶𝗼𝗻𝘀: I have created and published freely available video lectures on YouTube for courses in Bioinformatics, Image Processing, Data Mining & Data Science, and Computer Programming. • 𝗠𝘆 𝗩𝗶𝘀𝗶𝗼𝗻: Together we grow! • 𝗠𝘆 𝗠𝗶𝘀𝘀𝗶𝗼𝗻: To be the best for your business or product!

🔸 Certified AI & Machine Learning Engineer 🔸 LLM, AI Chatbots, RAG & Automation Systems 🔸 AI Developer | Python | React.js | Node.js | AI Agent & Automation 🔸 End-to-End Ownership: Model → API → Frontend → Deployment 🔸 Post-Delivery Support Included I’m an AI and ML Engineer, designing and deploying end-to-end AI systems that solve complex real-world challenges. My background spans AI, Full Stack development, Computer Vision, NLP, LLMs, RAG architectures, and Multi AI Agents, enabling me to build intelligent, scalable, and automation-ready platforms for global clients. ✔️ 𝗪𝗵𝗮𝘁 𝗜 𝗕𝘂𝗶𝗹𝗱 AI & ML Systems • LLMs, GPT-powered apps, RAG pipelines, LangChain, LangGraph • Predictive analytics, recommendation engines, anomaly detection • NLP engines: sentiment analysis, chatbots (text/voice), summarization, custom transformers • Real-time object detection (YOLO, Detectron2), pose estimation (MediaPipe, OpenPose) Computer Vision Applications • Industrial defect detection (98%+ accuracy) • Worker-safety monitoring, multi-camera systems • OCR automation & image/video analytics • Real-time CV dashboards & monitoring tools Full-Stack AI Integration • Python: FastAPI, Flask, Django • Frontend apps: React, Next.js, Vue.js • AI microservices, scalable REST/GraphQL APIs • Deployments on AWS, GCP + edge devices (Jetson Nano, Raspberry Pi) AI Automation & Workflows • Social media automation (Instagram/TikTok/X) • E-commerce automation (Shopify/Woo) • Web automation & scraping (Selenium, Scrapy) • ML CI/CD pipelines (MLflow, Kubeflow, Docker, Kubernetes) ✔️ 𝗧𝗲𝗰𝗵 𝗦𝘁𝗮𝗰𝗸 AI & ML: Python, PyTorch, TensorFlow, Transformers, Scikit-Learn LLMs & Automation: GPT-4/5, RAG, LangChain, LangGraph, CrewAI Computer Vision: YOLOv8, OpenCV, Detectron2, MediaPipe Backend: FastAPI, Flask, Django, Node.js Frontend: React, Next.js, Vue.js Databases: MongoDB, PostgreSQL, MySQL, Firebase, Supabase DevOps: Docker, Kubernetes, MLflow, CI/CD Cloud & Edge: AWS, GCP, Azure, NVIDIA Jetson, Raspberry Pi ✔️ 𝗪𝗵𝘆 𝗪𝗼𝗿𝗸 𝗪𝗶𝘁𝗵 𝗠𝗲 • I ship production-ready AI solutions, not just prototypes • I bridge AI models with full-stack applications seamlessly • Transparent communication, clean documentation, and on-time delivery • I focus on real business impact, not unnecessary complexity 🧠 𝗥𝗲𝗮𝗱𝘆 𝘁𝗼 𝗯𝘂𝗶𝗹𝗱 𝗮 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻-𝗴𝗿𝗮𝗱𝗲 𝗔𝗜 𝘀𝗼𝗹𝘂𝘁𝗶𝗼𝗻? Let’s connect and bring your vision to life today.

I'm a Senior Data Scientist who has worked on a range of AI, Machine Learning, and Cybersecurity projects including deep learning, forecasting, recurrent and convolutional neural networks, financial prediction tools, natural language processing models and much more. In addition, I also have full-stack development experience working with customers worldwide. I'm knowledgeable in technologies such as Python, Tensorflow, Pytorch, Scikit-Learn, Keras, JavaScript, ReactJS, NodeJS, Pandas Numpy, SQL, and much more. I have helped build AI projects in a range of sectors, web applications for startups, implemented DS models for e-commerce leading to higher sales performance and retention initiatives along with overall DS and AI related options for businesses of all sizes.

About Me: I’m an AWS-certified Machine Learning Engineer with 4+ years of hands-on experience building and deploying production-grade AI and ML systems. I specialize in Large Language Model (LLM) development, Retrieval-Augmented Generation (RAG) pipelines, and Natural Language Processing (NLP) applications such as classification, summarization, and sentiment analysis. With deep expertise in AWS SageMaker and Bedrock, I help organizations move generative AI projects from proof-of-concept to production, ensuring scalability, compliance, and performance. My focus is on building reliable, maintainable AI solutions that integrate seamlessly into existing business workflows. Currently pursuing my MSc in Computer Science (AI/ML) at Georgia Tech, I combine strong theoretical foundations with proven practical skills to deliver solutions that truly drive business outcomes. Services I Offer: 1. Fine-tuning Large Language Models (LLMs) for domain-specific applications 2. Building end-to-end RAG (Retrieval-Augmented Generation) systems with vector databases (Pinecone, Bedrock KB, Opensearch ) 3. Developing AI Agents for intelligent automation and knowledge retrieval 4. Designing and deploying NLP solutions – classification, summarization, sentiment analysis, entity extraction 5. Prompt engineering and model optimization for improved performance and accuracy 6. Production ML Systems – CI/CD pipeline automation, model versioning, and continuous monitoring using AWS SageMaker and related services Skills: Programming: Python, PySpark, JavaScript, FastAPI AI/ML: LLM Development (Hugging Face, LangChain, Bedrock), NLP (spaCy, Transformers) Frameworks/Libraries: Scikit-learn, Pandas, NumPy, TensorFlow, PyTorch Cloud: AWS (SageMaker, Bedrock, Lambda, Glue, API Gateway, CloudWatch, S3) Model Lifecycle: MLflow, CI/CD for ML, Model Monitoring Databases: SQL, Pinecone (Vector DB), DynamoDB AI Optimization: Prompt Engineering, Model Fine-Tuning, Quantization Visualization: Tableau, AWS QuickSight Education: 🎓 MSc in Computer Science (AI/ML) – Georgia Institute of Technology, Atlanta, United States. 🎓 MSc in Cloud Computing (Software Engineering) – Munster Technology University, Cork, Ireland 🎓 BEng in Computer Engineering – Federal University of Technology, Owerri, Nigeria. Certifications: ✅ AWS Certified Solutions Architect – Associate ✅ AWS Certified Machine Learning – Specialty ✅ AWS Certified GenAI Developer – Professional (In progress) Industry Experience: AWS Data/ML Engineer @ HiveTekCorp (Dec 2025 - till present) MLOps Engineer @ Veterans United (Sept 2022 - Sept 2023) Data Scientist @ Veterans United (Dec 2020 - Sept 2022)

I'm a Python and C++ developer with 18 years of experience. I can efficiently handle a wide variety of tasks in data science, machine learning, software architecture and debugging, and more. I'm good at writing, and always prioritize clear communication in my work. I've recently worked on a project where I developed and optimized an OpenCV+YOLO object detection pipeline in C++ on multiple real-time video streams, which is currently in production at several customer sites. I'm particularly interested in speech and audio processing, and have worked on many projects in this domain. After graduating with a dual Bachelor's Degree in 2004, I was the lead developer of a product for distributed entity resolution and name matching at a Washington, DC area contractor for 6 years. Starting with an existing prototype and legacy C++ libraries, I combined Python and C++ to create a high-performance, distributed tool for use by data analysts. I documented the code and created automated tests, then trained several new developers who joined the team. I wrote our user documentation, and after our first release to customers, I worked with them closely to add new features and fix bugs. In writing and maintaining a high-performance distributed application I became proficient with the POSIX API including threading, the file system, and networking. Since leaving this contractor, I have been developing my own software product for foreign language learning. It takes user input from the microphone, and uses machine learning to present appropriate study material. The product is still in development. The prototype uses Python with NumPy, SciPy, Matplotlib, and Torch, and I have become proficient in using these tools to process and explore data and develop machine learning models. I also developed a speech corpus from public domain data, aligned the transcriptions with the audio using Kaldi, and extracted spectral features to be used in my speech processing pipeline. My dual Bachelor's is in Computer Science, and International Languages and Cultures - French. I studied computer science in Bordeaux, France for one semester and speak and write French fluently. My native language is American English. I also studied Chinese for 2 years in college, and taught English in China for one year after graduation. I speak Mandarin fluently and read and write Chinese. I have done professional translation work from both French and Chinese to English, including a translation of Xu Dan's book titled Introduction to Chinese Syntax, published by l'Asiathèque. I have done editing and translation work for Xu Dan and other linguists on many projects over the past 18 years. In both this and my software development work, I take pride in producing high-quality work.