Anurag Kotla

Experience

• 

  Infosys

  Jan 2021 - Aug 2022

  Data Analyst

  • Expertly employed Python for data manipulation, statistical analysis, and predictive modeling, vital in fostering innovative AI-driven solutions, driving advancements in technology and problem-solving within the organization.• Utilized SQL and MySQL databases proficiently to query and manage extensive datasets, guaranteeing data integrity and accessibility for analysis and modeling purposes, enhancing efficiency and accuracy in decision-making processes.• Implemented advanced Excel features such as pivot tables, VLOOKUP, INDEX-MATCH, and array formulas, applying calculus concepts to optimize data analysis and reporting, resulting in a 30% reduction in data processing time.• Implemented various machine learning algorithms such as Regression Analysis, Clustering, Decision Trees, and Neural Networks in data analytics projects to extract actionable insights from complex datasets, driving data-informed decision-making processes.• Designed and developed interactive data visualizations using tools like Matplotlib, Seaborn, Tableau, and Power BI, facilitating the communication of key findings and insights to stakeholders. Show less

• 

  VEN SOLUTIONS LLC

  Nov 2023 - Oct 2024

  Data Scientist

  • Developed and deployed Advanced AI/ML algorithms, enhancing predictive analytics accuracy by 20% and reducing infrastructure costs by 15% for cloud data solutions.• Led cross-functional teams in designing and implementing scalable data pipelines using Apache Spark and AWS Glue, enabling real-time insights and integration of disparate data sources for enhanced data analytics.• Created and maintained custom machine learning models with TensorFlow and sci-kit-learn, driving a 25% increase in customer engagement and retention rates through actionable insights into customer behavior.• Utilized Hugging Face's state-of-the-art transformer models, such as BERT and GPT, to develop cutting-edge NLP solutions for sentiment analysis, text summarization and named entity recognition, enhancing Intex's ability to derive actionable insights from textual data and improve customer support automation efficiency by 25%.• Implemented LLMS on Google Gemini Pro, analyzing user behavior and ad data, and enhancing ad targeting algorithms. Achieved a 25% rise in click-through rates and boosted ROI by 20% for advertising campaigns.• Conducted exploratory data analysis and feature engineering techniques to uncover trends in large datasets, providing valuable insights for strategic decision-making and product roadmap prioritization• Conducted comprehensive exploratory data analysis (EDA) on datasets including sales, coolers, and weather data (318,672 rows, 8 columns, no null values).• Engineered advanced features such as interaction variables and rolling statistics to enhance trend and seasonality capture.• Applied Recursive Feature Elimination (RFE) for feature selection to improve model efficiency. Show less

• 

  Oraczen

  Nov 2024 - now

  Generative AI Engineer

  • Utilized advanced vector databases such as ChromaDB and Pinecone, enhancing AI application data retrieval speeds by 60%, substantially improving overall performance and efficiency.• Developed scalable GenAI solutions with LangChain and LlamaIndex frameworks, resulting in a notable 75% decrease in model training duration, optimizing efficiency and enhancing project productivity significantly.• Tailored and refined open-source and proprietary LLM models, achieving a 50% boost in accuracy rates and a 40% reduction in inference time, enhancing overall model performance significantly.• Led deployment of GenAI models on AWS Bedrock, enabling a 70% improvement in application scalability and reliability, while leveraging cloud services to reduce operational costs by 30%.• Orchestrated creation of transformative GenAI applications, including an LLM App using Llama2, driving a 50% increase in user productivity and engagement metrics.• Designed data pipelines to integrate text and visual data with precise synchronization for fine-tuning GPT-4o.• Implemented embeddings using advanced vectorization techniques for text and images, improving multimodal search capabilities.• Leveraged MidJourney Python API for controlled design generation using HYDE techniques.• Fine-tuned image generation models (Stable Diffusion XL) with specialized jewelry datasets for high-accuracy outputs.• Developed advanced workflows for ComfyUI focused on image generation, inpainting, outpainting, and integrating ControlNet models, including Canny, M-LSD, HED Boundary, Scribble, and Human Pose, to achieve precise and creative outputs.• Achieved efficient storage and retrieval of multimodal data using vector embeddings.• Enhanced 3D modeling workflows through Hugging Face tools, enabling detailed visual outputs. Show less