Varun Badugu

AI Engineer with 4+ years of experience designing, developing, and deploying production-grade AI, machine learning, and Generative AI solutions. Experienced in building Retrieval-Augmented Generation (RAG) systems, AI agents with tool-calling capabilities, LLM-powered applications, prompt engineering and evaluation workflows, and LLM fine-tuning solutions. Strong background in Python, PyTorch, TensorFlow, transformer architectures, vector search, NLP, and scalable ML pipelines.

Proven experience optimizing model training and inference, deploying AI/ML workloads on AWS, and implementing MLOps practices using Docker, Kubernetes, CI/CD, and infrastructure automation. Focused on building reliable, scalable AI systems that improve accuracy, reduce latency and cost, and automate complex business workflows.

AI & ML Engineer - Meta - CA

(2024-09)

• Design and develop production-grade AI and machine learning pipelines using Python, PyTorch, and TensorFlow, improving distributed model-training efficiency by 40% through optimized data loading, caching, and multi-node training strategies.
• Architect and deploy transformer-based NLP solutions using BERT and RoBERTa for large-scale content understanding, achieving 92% model accuracy while reducing inference latency by 55% through model pruning and TensorRT optimization.
• Build Retrieval-Augmented Generation (RAG) solutions using LLaMA 3, FAISS, vector-based retrieval, and prompt engineering techniques, improving grounded-answer accuracy by 25% while reducing hallucinations through optimized retrieval and contextual grounding.
• Develop AI agent workflows with tool-calling capabilities to enable LLM-powered applications to interact with external systems, retrieve contextual information, and execute multi-step tasks.
• Implement LLM fine-tuning and prompt-engineering workflows to adapt foundation models for domain-specific use cases and improve response relevance, consistency, and task performance.
• Design prompt and LLM evaluation approaches to compare model responses, retrieval quality, and prompt variants, enabling systematic optimization of Generative AI application performance.
• Spearhead A/B testing frameworks for AI/ML deployments, implementing statistical validation and automated experiment reporting to reduce experimentation cycles by 40%.
• Design SQL-based experiment and hyperparameter tracking capabilities supporting 1,000+ model-training runs, enabling faster analysis and reducing model iteration time by 50%.
• Partner with Data Engineering and MLOps teams to productionize end-to-end ML systems, improve feature-store integration, and reduce model inference costs by 30%.
• Implement ML-focused CI/CD pipelines using GitHub Actions and internal tooling to automate model testing, validation, and deployment, reducing production rollout time by 65%.

AI & ML Engineer - Wipro - India

(2020-05 - 2023-07)

• Developed end-to-end machine learning solutions using Python, Scikit-learn, XGBoost, Random Forest, and neural networks for customer churn prediction, achieving 89% predictive accuracy and contributing to a 22% reduction in annual client attrition.
• Built scalable ML training and deployment pipelines on AWS SageMaker, AWS Lambda, and Step Functions, automating model workflows and reducing deployment time by 60% and cloud costs by 35%.
• Contributed to an AI-powered fraud-detection platform for a major banking client using anomaly-detection and predictive modeling techniques, reducing false positives by 40% and contributing to more than $1M in annual savings.
• Designed feature-engineering and data-processing workflows for large structured datasets using Python, SQL, and statistical analysis techniques to improve model reliability and predictive performance.
• Optimized PostgreSQL and MySQL databases supporting ML applications, improving data retrieval performance by 50% through query optimization, indexing, and stored procedures.
• Developed hybrid statistical and machine-learning solutions using Python, Scikit-learn, and R for financial risk analysis, improving predictive accuracy by 18% while reducing processing time by 30%.
• Built interactive analytics dashboards in Tableau to communicate model predictions, trends, and business KPIs to technical and non-technical stakeholders.

M.S. - Computer Science - University of Illinois at Springfield Springfield, IL

B.Tech. - Computer Science and Engineering - Mahaveer Institute of Science and Technology Hyderabad, India