Sai Anjali

About Sai Anjali

Machine Learning Engineer specializing in Generative AI and Large Language Model (LLM) systems, with 4+ years of experience designing scalable AI solutions and production-ready ML platforms across enterprise environments. Experienced in building Generative AI applications, LLM evaluation frameworks, and Retrieval-Augmented Generation (RAG) pipelines to improve model reasoning, reliability, and enterprise knowledge retrieval.🤖 Skilled in developing LLM evaluation pipelines, adversarial prompt testing systems, and automated benchmarking frameworks to detect hallucinations, bias, and safety risks in AI deployments. Built distributed machine learning systems and scalable experimentation pipelines using PyTorch, Hugging Face Transformers, Ray, and ML flow to accelerate large-scale model validation workflows. Designed LLMOps and MLOps workflows, including experiment tracking, model monitoring, versioning, and continuous evaluation pipelines for production AI systems. Experienced with cloud-native AI infrastructure on AWS, deploying containerized ML services using Docker, Kubernetes, GPU inference pipelines, and scalable microservices architectures.🧠 Hands-on expertise in Transformer architectures, RLHF pipelines, reward modeling, and prompt engineering to improve LLM reasoning quality and task completion accuracy. Built RAG-based knowledge retrieval systems integrating vector databases (FAISS / Pinecone), semantic embeddings, and large-scale document intelligence pipelines. Focused on Responsible AI and model safety, implementing hallucination detection, bias monitoring, interpretability techniques, and adversarial testing for robust AI deployment.🤝 Passionate about building scalable AI platforms and intelligent systems that bridge research and production while collaborating with ML researchers, engineers, and product teams.

Experience

1. Machine Learning Engineer

   Scale AI

   May 2025 — Present · San Francisco, CA, US

   Designed automated evaluation pipelines to benchmark large language models (LLMs) for hallucination, reasoning accuracy, and safety compliance, improving model reliability by 32% across enterprise AI deployments.• Built adversarial prompt testing framework and synthetic benchmark datasets to stress-test generative AI models, reducing critical safety failures by 27% before production releases.• Developed scalable evaluation workflows using distributed AI systems, compute architectures, and automated scoring pipelines, accelerating model validation cycles by 40% and improving continuous quality monitoring.• Implemented LLM-as-a-Judge evaluation techniques using PyTorch and transformer models to automatically score reasoning outputs, improving automated model assessment accuracy and experimentation efficiency.• Engineered high-quality evaluation datasets through prompt engineering, synthetic data generation, and human-in the-loop feedback pipelines while collaborating with cross-functional research and product teams.• Built machine learning experimentation pipelines leveraging Python, HuggingFace Transformers, Ray distributed computing, and MLflow experiment tracking to support large-scale model testing workflows.• Strengthened AI safety through adversarial testing, bias detection, hallucination monitoring, and interpretability techniques while clearly communicating insights across engineering, research, and leadership teams.• Designed scalable data processing pipelines using Python, Spark, and Kubernetes orchestration while deploying evaluation services on AWS to support high-volume automated testing workloads.• Developed real-time analytics dashboards for model benchmarking and evaluation insights using cloud-native data pipelines and monitoring services deployed within AWS infrastructure.

Education

• Saint Louis University

  Master's degree, Master of Science