Azema Shaik

Machine Learning Engineer:

• Build and maintain practical and reliable large language model systems, collaborating with cross-functional teams to fine-tune, evaluate, and integrate models into real-world business workflows.
• Fine-tuned LLaMA 3.1 8B models using USloth and LoRA techniques, then exported optimized checkpoints to Ollama for seamless local deployment.
• Designed and implemented end-to-end evaluation pipelines to test model outputs for toxicity and bias, ensuring that every release met our ethical and performance standards.
• Diagnosed and resolved GPU out-of-memory challenges during training and inference, adjusting batch sizes, mixed-precision settings, and memory-efficient libraries to keep projects on schedule.
• Collaborated with the legal team to leverage LLMs for ranking Master Service Agreement (MSA) clauses, making contract review faster and clearer for non-technical stakeholders.
• Automated data collection, scoring, and reporting steps within the evaluation pipeline to minimize manual review and accelerate model iteration cycles.
• Monitored model performance metrics in production, identified areas for improvement, and iterated on prompt strategies to maintain high accuracy and relevance.
• Wrote clear documentation and created usage examples so that other teams (legal, compliance, product) could use our tools without deep AI expertise.

This role required hands-on problem solving, from customizing fine-tuning workflows to troubleshooting GPU memory issues, all while keeping ethical considerations top of mind.

DevOps Engineer:

• Developed and maintained Jenkins CI/CD pipelines to automate the build, test and deployment processes.
• Configured and managed Jenkins jobs to ensure efficient and reliable continuous integration and delivery.
• Deployed applications as microservices using Docker, Kubernetes, Helm ensuring scalability, reliability and efficient resource utilization.
• Monitored and troubleshoot CI/CD pipeline to ensure smooth and continuous delivery of applications.

• Designed and implemented a logging system in Go, inspired by Python’s logging module.
• Supported multiple log levels (info, warning, error, debug), with easy control over log verbosity.
• Enabled formatted log output with timestamps, log levels, and custom message templates.
• Allowed output to different destinations (console, file) for flexible logging in CLI tools and backend services.
• Built custom handlers to manage log message routing and formatting.
• Made logging configuration simple and intuitive, similar to Python’s style.
• Ensured thread-safe logging to handle concurrent operations.
• Added clear documentation and example usage to make the logger easy to adopt in Go projects.
• Used this project to deepen understanding of Go’s interfaces, concurrency, and error handling.
• Applied the logger in personal automation tools to monitor execution and debug efficiently.

• Developed a revolutionary car design platform by harnessing the capabilities of the Stable Diffusion model to create an intuitive design tool for car enthusiasts.
• Integrated user-friendly interfaces to facilitate easy input of design
  preferences and receive AI-generated visual outputs.
• Utilized Stable Diffusion's diffusion techniques to ensure the generation of high-quality, realistic car designs.
• Demonstrated expertise in adapting the Stable Diffusion model to cater to the specific requirements of the automotive customization domain.
• This solution was chosen to be presented at the Google Next Summit within my current organization

• Developed GenZBot, a feature-rich AI-powered virtual assistant, using stateof-the-art technologies.
• Integrated DALL-E for text-to-image generation, enabling digital art creation and visual responses to prompts.
• Utilized Whisper for accurate speech recognition, enhancing hands-free user interactions.
• Employed ChatGPT to enable natural language understanding and
  generation for intuitive user conversations.
• Integrated LangChain to optimize the integration of various language
  models, accelerating development.
• Implemented real-time internet connectivity for up-to-date information retrieval and web-related tasks.

• Designed and developed an innovative essay-writing app, utilizing Large Language Model technology to facilitate content creation.
• Integrated AI modules to synthesize information and generate essay content based on user-input questions.
• Implemented summarization features to distill key points, enabling users to refine their ideas and arguments.
• Developed a quiz question generator, allowing users to self-assess their understandingand identify learning gaps.
• Created an intuitive user interface for seamless input, customization, and output of essays and quizzes.

• Designed and implemented a personal project that translates human language queries into SQL queries using GPT-3.5.
• Built an interactive interface in Streamlit to allow users to enter questions and receive corresponding SQL queries.
• Integrated GPT-3.5 via API to automate the process of generating SQL from natural language.
• Experimented with prompt engineering and input validation to enhance the reliability and relevance of generated queries.
• Worked with JSON for data handling between user interface and the language model.
• Tested and refined the tool to support a variety of user requests and common database scenarios.
• Automated manual SQL writing tasks, making database access easier for those with little or no SQL knowledge.
• Documented the project with clear usage instructions and practical examples for others to follow.
• Gathered feedback from users to further improve functionality and user experience.
• Used this project to explore and deepen my skills in generative AI, user interface development, and workflow automation.

Let me know if you want a shorter or more detailed version, or if you want the focus shifted!

• Researched and resolved issues regarding integrity of data flow into databases.
• Identified and documented detailed business rules and use cases based on requirements analysis.
• Created various Excel documents to assist with pulling metrics data and presenting information to stakeholders for concise explanations of best placement for needed resources.