Masam Varshith Yadav

Passionate Generative AI / Agentic AI Engineer with strong foundations in Python and modern LLM frameworks. Experienced in building multi-agent systems with LangGraph and RAG pipelines, and scalable AI applications using LangGraph and LangChain. Interested in developing production-grade autonomous AI systems that integrate LLMs, tools, and real-world data.

Open-Source Contributor at Intel OpenVINO Toolkit (NNCF) (2025-01 – Present)

Contributed to widely used open-source machine learning libraries, improving model architectures, data preprocessing utilities, and neural network compression workflows.

• Contributed PR #3759 improving checkpoint loading and weight-mapping stability in neural network compression workflows
• Implemented support for data-aware weight compression in MatMul operations (transpose_b=False)
• Updated test coverage

Open-Source Contributor at skrub (2025-01 – Present)

Contributed to open-source data processing library.

• Contributed PR #1910 improving numeric data handling in the Cleaner module
• Introduced explicit behavior modes for numeric_dtype (passthrough, parse, float32)
• Enabled controlled parsing of numeric-like strings while preserving existing TableVectorizer behavior

Open-Source Contributor at PyTorch Forecasting (2025-01 – Present)

Contributed to time-series forecasting library.

• Contributed PR #2041 implementing an encoder-decoder LSTM model with native multi-target forecasting support
• Compatible with AutoRegressiveBaseModel and MultiLoss
• Full decoder-horizon predictions and test coverage

Bachelor of Technology (B.Tech) in Computer Science – Polaris School of Technology (2024-01)