I'm a Data Scientist (3+ yrs) experienced in building and deploying AI solutions that drive measurable business value.
Currently working at CDM Smith (Engineering & Construction) where I'm building Generative AI systems, orchestrating LLM pipelines using LangGraph, and developing RAG-based internal tools to automate schedule generation and knowledge retrieval.

I enjoy working at the intersection of Machine Learning, MLOps, and Scalable Systems, transforming ideas into end-to-end production-ready solutions.

• Generative AI: RAG, LangChain, LangGraph, Prompt Engineering, Azure OpenAI
• Machine Learning: Supervised/Unsupervised, NLP, CNNs, Time Series, XAI (SHAP, LIME)
• MLOps: FastAPI, Docker, MLflow, GitHub Actions, CI/CD
• Data Engineering: Snowflake, BigQuery, Azure, AWS
• Visualization & BI: Power BI, Tableau, Matplotlib, Seaborn
• Languages: Python, SQL, Bash
• Frameworks: TensorFlow, PyTorch, Scikit-learn, Pyspark

• Built a RAG-based LLM platform reducing schedule generation time from 30 days - 5 minutes.
• Developed FastAPI backend, integrated Azure OpenAI models (GPT-4.1, 4o, o3) with stateful workflows via LangGraph.
• Added model comparison, intent detection, and Azure Application Insights logging.

• Automated analysis of 50+ competitors using Azure Document Intelligence and Power BI.
• Parsed PDFs and XBRL filings with Python to generate actionable insights.

• Achieved 0.95 F1-score using LightGBM with EDA, feature engineering, and model tuning.
• Built CI/CD pipelines via GitHub Actions, deployed with Flask, Docker, AWS ECS, and Postman validation.
  View Project

• Expanding LangGraph expertise for agentic orchestration of LLM pipelines.
• Building domain-specific LLM evaluation frameworks (BLEU, ROUGE, METEOR, BERTScore).
• Exploring autonomous LLM agents for knowledge extraction and workflow automation.

ashutoshsinha519@gmail.com
LinkedIn
Portfolio / Projects

"Turning data into decisions and AI into impact."