Jatin Chandani

Built a property analytics platform using computer vision and generative AI, automating report generation and improving operational efficiency by 40%. Engineered and fine-tuned transformer-based models for information extraction and summarization, boosting accuracy by 25%. Developed document processing pipelines with PDF ingestion, embedding generation, and FAISS-based semantic search to enhance retrieval relevance. Integrated text-to-SQL models, increasing data accessibility for non-technical users. Collaborated with stakeholders and cross-functional teams to identify pain points and deliver AI-driven insights. Deployed a modular Streamlit dashboard, accelerating decision-making by 30%. Contributed clean Python code to a shared GitHub repo and participated in Agile sprints improving delivery timelines by 15%.

Built a property analytics platform leveraging computer vision and generative AI, automating report generation to improve operational efficiency by 40%. Engineered and fine-tuned transformer-based models for information extraction and summarization, boosting accuracy by 25%. Developed document processing pipelines involving PDF ingestion, embedding generation, and FAISS-based semantic search to improve retrieval relevance. Integrated text-to-SQL models for natural language querying, enhancing data accessibility for non-technical users. Worked closely with business stakeholders to identify automation opportunities and deployed a Streamlit dashboard to visualize model outputs, accelerating decision-making by 30%. Contributed production-ready Python code in a shared GitHub repository and participated in Agile sprints to improve delivery timelines by 15%.

Designed and deployed ML models for property analytics using supervised and unsupervised techniques. Fine-tuned transformer models such as BERT for text summarization achieving a 25% accuracy increase. Engineered semantic search pipelines using document embeddings and FAISS, improving data retrieval speed and relevance by 35%. Conducted A/B testing for model evaluation. Automated ETL workflows using Airflow and Luigi, reducing manual processing by 35%. Deployed scalable applications via Streamlit, FastAPI, and Vertex AI on GCP with integrated performance monitoring. Collaborated with product and engineering teams and implemented responsible AI practices. Expanded big data capabilities using Hadoop, Spark, and Dataproc for production-scale processing.