Anant Pratap Shahi

About Anant Pratap Shahi

Data Scientist & Machine Learning Engineer with experience building and deploying AI systems using LLMs, agentic frameworks, and scalable ML pipelines. My work bridges data science, software engineering, and AI product development—driving real-world impact across industries.I\'ve delivered end-to-end ML solutions, fine-tuned large language models (LLMs), and designed intelligent agents for automation and decision-making at scale. From prompt engineering to model deployment and evaluation, I operate across the full AI lifecycle.My recent work involves:• Implementing LLM-powered agents using LLM Frameworks • Building modular ML pipelines with Python, TensorFlow, and PyTorch• Developing production-ready AI workflows with focus on performance, interpretability & ethics Currently open to roles in: Data Science (ML, NLP, LLMs) AI Engineering Applied AI & Agentic Systems🧠 Tech Stack & Tools: Python, PyTorch, TensorFlow, LangChain, OpenAI, HuggingFace, Scikit-learn, Pandas, SQL, Docker, Git, REST APIs Actively exploring opportunities with AI-driven teams at Fortune 500s and high-impact startups. Let’s connect if you\'re hiring or building something game-changing.

Experience

1. Data Scientist - Product Data Solutions

   S&P Global

   Sep 2024 — Present · Gurugram, IN

   International Trade Reporting QA & Anomaly Detection -Spearheaded the development of NLP-powered quality assurance pipelines for international trade reporting, utilizing LLMs with Python-enforced custom rule sets written in layman-friendly language-Designed and deployed an end-to-end anomaly detection service, enabling proactive trade validation prior to reporting, significantly reducing trade rejection rates-Delivered measurable improvements in reporting integrity and compliance with global regulatory standards.Advanced Query Engine for Corporate Actions -Engineered a hybrid intelligent query system using LLMs, Langchain, Guardrails, Elastic DB, and Python/Java microservices to handle complex client inquiries on corporate actions-Achieved query response time reduction from several hours to under 5 seconds, with <5s average latency-Designed for scalability with plug-and-play integration across multiple enterprise products; enabled multilingual querying for global client accessibility.Smart Summarisation Engine – Simplified Narrative -Developed and deployed a multilingual summarisation engine on AWS using LLMs, Langchain, and Guardrails to translate complex corporate action data into concise, chronological, and insight-driven narratives-Enabled stakeholders to process high-volume information in seconds with actionable takeaways in their preferred language.Depositary Service Fee Feed – New Revenue Line -Conceptualised and built a full-stack service to track, aggregate, and distribute depositary service fees from diverse sources-Successfully launched a new revenue-generating product offering by automating an otherwise manual, fragmented process.

Education

• MIT ADT University

  Bachelors in technology, Computer Science

  2017 — 2021