- Led the development of advanced chatbot solutions for clients like SAP and Toyota, focusing on automating and improving user interactions through the application of Large Language Models (LLMs).- Increased the Mean Reciprocal Rank (MRR) by over 30% and the Hit Rate by 50% in a Retrieval-Augmented Generation (RAG) system, significantly enhancing information retrieval accuracy. This improvement resulted in more precise answers, greatly boosting client satisfaction with the search functionality.- Streamlined processes within the machine learning team by managing MLOps to improve efficiency and productivity. Led system and architecture design, optimized existing projects/products, and implemented best practices and structured workflows. Managed the full lifecycle of ML projects, from planning and development to deployment and monitoring, utilizing AWS.- Spearheaded the design and implementation of a microservices architecture for a RAG (Retrieval-Augmented Generation) system on AWS, ensuring both scalability and reliability. Successfully migrated the system to the client's AWS environment with minimal disruption.- Deployed open-source Large Language Models (LLMs) like Llama 2 using vLLM on AWS EC2, scaling the system to support hundreds of concurrent users while ensuring low latency and high throughput for a seamless user experience.- Led deployment efforts for the ML team by containerizing all projects with Docker, establishing CI/CD pipelines using GitHub Actions, configuring NGINX with custom domains and TLS, automating infrastructure setup with Python and AWS CDK, and deploying on AWS for optimal scalability, reliability, and cost-efficiency.

Role Description:As a Machine Learning Engineer at Vacon.ai, I leveraged cutting-edge models, conducted experiments, and integrated them into production environments for real-world impact. My work covered computer vision and NLP projects, showcasing my versatility and expertise in machine learning.Key Projects:Computer Vision Projects: - Virtual Renovation Project: I developed a virtual renovation system for visualizing house renovations using Stable Diffusion for realistic images, SAM for precise segmentation and ControlNet for detailed control. This tool aided architects, interior designers and homeowners in accurately and creatively visualizing design options. - Virtual Try-On System: I developed a virtual try-on system for fashion and retail using LORA for Stable Diffusion, enabling users to see how clothing items looked on them. The system, thoroughly tested for realism and efficiency, enhanced customer engagement and satisfaction with an immersive shopping experience. - Object Detection Systems: I developed a zero-shot object detection system using Grounding DINO and CLIP for identifying objects without prior training and implemented an efficient object detection project using YOLO for accurate feature localization. - Facial Recognition System: I developed a facial recognition system employing AWS Rekognition to improve security and access control, ensuring robust and reliable identification.NLP Projects: - Document Retrieval System: I engineered a document retrieval system using the Haystack model, which efficiently retrieved relevant documents from extensive corpora in response to user queries, enhancing information accessibility. - Recruitment Streamlining: I developed a machine learning-driven system for extracting education, job experience and skills from resumes, enabling the identification of top candidates and streamlining the recruitment process.#StableDiffusion #Midjourney #LLM #VirtualRenovation #VirtualTryOn

As a Full Stack Web Developer at Cipher Coders, I specialized in building dynamic, scalable applications using React.js for the front-end and Node.js with Express.js for the back-end. I developed end-to-end solutions, from creating intuitive UIs to building robust server-side logic, ensuring seamless user experiences and efficient data handling.Key Projects: 1. Blog Platform with Admin Panel I developed a full-featured blog platform where users could read, comment and share posts. The admin panel allowed authenticated users to create, manage and delete content. Using React.js for the front-end and Node.js/Express.js for the back-end, I integrated MongoDB for content storage and implemented features like authentication, input validation, and secure data handling. 2. E-commerce Store with Product Management I worked on an e-commerce platform for managing products, processing orders, and providing a seamless shopping experience. The front-end, built with React.js, allowed users to browse and purchase products, while the back-end in Node.js/Express.js handled user authentication, product inventory and order processing. I integrated Stripe for secure payment handling. My responsibilities included managing ongoing development, adding new features and optimizing the platform’s performance and user experience.Responsibilities: 1. Designed and developed scalable full-stack applications. 2. Collaborated with cross-functional teams to deliver high-quality solutions. 3. Maintained and enhanced existing applications, adding new features and functionalities. 4. Ensured security, performance, and scalability in all aspects of the applications.In this role, I gained hands-on experience in both front-end and back-end development and contributed to impactful projects that provided real value to users and businesses.#FullStackDevelopment #ReactJS #NodeJS #WebDevelopment #Ecommerce #BlogPlatform #Tech