Sasank Dhulipudi

· Automated auto trade reconciliation by extracting information from unstructured chat confirmations and built classification models for trade instruments, reducing manual effort by 80%.

· Developed a retrieval‑augmented generation (RAG) assistant using large language models and embeddings; rigorous evaluation improved user efficiency by 50% and enabled decommissioning of legacy tools worth approximately $800K.

· Developed Standard Operating Procedure GPT assistant to streamline SOP-related query responses, enhancing procedural adherence and knowledge accessibility across teams.

· Implemented OCR data extraction pipelines from PDFs using ABBYY and Azure services, facilitating efficient document ingestion.

· Analyzed automotive sensor data and developed predictive models to identify potential system failures, enabling proactive maintenance and improving safety.

· Performed exploratory data analysis and interactive visualizations to communicate insights from sensor data to stakeholders.

· Applied clustering techniques to identify accident hotspots and inform safety improvements for vehicle platforms.

· Developed multilevel classification system for medical reports, reducing disease detection time and diagnostic costs.

· Designed and trained NLP models for processing medical reports, including data collection, cleaning, tokenization, stemming, lemmatization, and classification using SVM and Random Forest.

· Built and evaluated models using TF‑IDF, Word2Vec, and Bag‑of‑Words; monitored sensitivity and accuracy, comparing models with ROC curves.

· Automated repetitive tasks using robotic process automation, developing DLLs and scripts that reduced manual effort and saved approximately $500K annually.

Deployed and monitored automation workflows, ensuring system reliability and optimal performance.

Developed and maintained Windows and web applications for internal and client‑facing solutions