I’m a Machine Learning Engineer who believes the real magic isn’t in training a model on a GPU cluster, but in making sure it works reliably for real people, in messy, unpredictable real-world environments.
Got my B.Tech and M.S. by Research from IIIT Hyderabad, where I learned that research papers and production code speak very different languages. These days I'm working with LLMs, RAG systems, and Document Intelligence. Sometimes it even works on the first try.
IIIT Hyderabad (Aug 2017 - Dec 2022)
B.Tech & M.S. by Research in Electronics and Communication Engineering
CGPA: 8.0/10
Published Work:
📄 Structural and Functional Pathology in Cocaine Use Disorder Progress in Neuro-Psychopharmacology and Biological Psychiatry (2024) J. Rasgado-Toledo, S. S. Duvvada, et al. A multimodal fusion approach to understanding structural-functional pathology in substance use disorders
- 🤖 Large Language Models: RAG, MCP Protocol, Fine-tuning, Prompt Engineering
- 🔍 NLP & Information Retrieval: Elasticsearch, Haystack, LangChain, LangGraph
- 👁️ Computer Vision: OpenCV, MediaPipe, Object Detection
- 📊 Deep Learning: GANs, Transformers, Hawkes Processes
- 📈 Signal Processing: PPG Analysis, Time-Series Forecasting
- Big Data Platforms: Databricks, Delta Live Tables, Medallion Architecture
- Vector Databases: Milvus, ChromaDB, FAISS
- Data Engineering: PySpark, NumPy, Pandas, SQLModel
- Databases: PostgreSQL, SQL Server
- MLOps Tools: MLflow, Databricks Asset Bundles, Docker, fastAPI
- Deployment: Docker, REST APIs, CI/CD Pipelines
- Building RAG systems that don't hallucinate medical advice (harder than it sounds)
- Making large-scale data pipelines that actually finish running before the heat death of the universe
- Contributing to open-source when I'm not fighting with Docker containers
- Figuring out how to make vector databases and SQL databases be friends
- Trying to convince people that "just prompt ChatGPT" isn't always the solution
Always up for chatting about ML projects, debugging production nightmares, or why your model works on your laptop but nowhere else.
- 📧 Hit me up for collabs, opportunities, or hot takes on the latest AI hype
Sometimes, I still sit there just watching logits unfold, token by token, wondering what the model really understands. Not because I need to fix it, but because I want to know.