sujan ks

At FlyRankAI, I developed machine learning baselines to identify webpages at risk of search-performance decline using historical search and click-performance data. I built a content-refresh opportunity scoring and ranking pipeline that achieved 0.680 Precision@50.

I also performed data validation, feature analysis, leakage checks, and signal auditing, then conducted Top-20 error analysis to support data-driven content-refresh decisions.

At AyushLab (isiri technologies pvt ltd ), I developed a hybrid AI-based invoice-processing pipeline combining EasyOCR, NLP preprocessing, regular expressions, rule-based extraction, and an LLM-based fallback. I built structured JSON extraction and validation workflows and developed REST APIs using FastAPI.

In my projects, I developed GeoSentinel, a YOLOv8 segmentation-based landslide detection system, and a RAG-powered portfolio assistant using BAAI/bge-m3 embeddings and Supabase pgvector. I’m studying Artificial Intelligence and Machine Learning at N.M.A.M. Institute of Technology.