Nikhil Vajramatti

I’m an ML Engineer with 3+ years of hands-on experience taking models from notebook to production. My work spans geospatial computer vision (U-Net rooftop segmentation), time-series price forecasting, and NLP/RAG for company/loan relationship extraction. I care about clean APIs, CI/CD, and writing runbooks that others can trust.

ML Engineer with 3+ years of hands-on experience shipping end-to-end ML products across geospatial computer vision, time-series forecasting, and NLP/RAG. Built a U-Net pipeline to segment industrial rooftops from satellite tiles (GeoPandas/Solaris/Mapbox), LSTM+CNN models for electricity price direction, and relation-extraction services that map company/loan links from SEC filings and news into Neo4j graphs. Comfortable owning the stack: data ingestion (Airflow), model training/fine-tuning (PyTorch; Llama LoRA/QLoRA), evaluation, and serving (FastAPI, Docker, GitLab CI/CD) with PostgreSQL persistence. Reduced manual website discovery by ~60% through a company-site analyzer using a fine-tuned Llama model. Mentor interns, create runbooks, and drive reliable releases.

Bachelor of Engineering, Mechanical Engineering at Dayanand Sagar College of Engineering