Gnaneswar Prakash K

Prompt Engineering: Skilled in crafting clear, precise, and contextually relevant prompts to fine-tune/optimize the output and performance of AI language models. LLM Evaluation: Expertise in reviewing and ranking LLM-generated code using Jupyter notebook, and natural language responses, providing detailed explanations for evaluations. Data Analysis: Skilled in Python programming to solve complex tasks, proficiently analyze real-world messy datasets, formulating hypotheses and conducting data-driven tests using to train/evaluate AI models. Model Response Crafting: Adept at creating ideal model responses and articulating the criteria for high-quality responses.

Created detailed insights using time series analysis to support strategic decision-making for various NHS trusts and private sector healthcare providers based on monthly, quarterly and yearly data. Implemented data visualization techniques using matplotlib and created highly effective PowerBI dashboards, resulting in the generation of actionable insights for the business. Automated the process of creating and preprocessing datasets using SQL and VBA programming in Excel, resulting in accuracy, reliability, and reduction of 20 weekly manual hours.

Conducted extensive data collection from Covid-19 testing sites and in-depth analysis of data collected, including key data points such as trends in the number of tests booked throughout the year, successful tests versus void tests, consent management of medical data and data anonymization of patient health data. Implemented regression model in Python using scikit-learn and TensorFlow for analysis and prediction of the future patient footprint to reduce strain on the healthcare system and deployed using AWS SageMaker. Identified the number of true positives, true negatives and type-I and type-II errors in order to determine the accuracy and precision of lateral flow test compared to the RT-PCR test to optimize monetary spend for the NHS.

Created and documented efficient Python scripts with 90% lesser memory footprint to derive insights from unstructured textual documents through our personalized cutting-edge Al solutions involving Bag of Words model and OCR. The lower memory footprint reduced server costs. Spearheaded the creation of the testing team to enable automated testing and reduce bugs in the pipeline, resulting in improved accuracy and reliability that contributed to a 22% increase in revenue year over year. Implemented NLP to summarize legal documents provided by the clients; conducted monthly 1:1 meetings to resolve issues and to ensure they understood the technical aspects of the solution provided in business context.

Worked with a research team of 3 and created a dataset for the research topic Detecting sub circuits used in a complex analog circuit using Machine Learning. Prepared complete documentation of research activities for reference purposes and evaluated existing research techniques to recommend improvements to meet tight deadlines for projects.