JIJIN A V

Data Scientist with over 9 years of experience in building and maintaining enterprise-grade conversational AI solutions. Extensive expertise in Google Dialogflow and Google Cloud Platform, including BigQuery, Cloud Functions, and scalable cloud-native architectures. Proven track record in delivering automation solutions that enhance user experience and operational efficiency. Hands-on experience with Agentic AI, having designed and implemented an AutoGen-based multi-agent conversational system enabling autonomous task planning, tool orchestration, and inter-agent collaboration. Skilled in designing KPI dashboards and operational monitoring solutions using Google Looker Studio to drive data-driven decision-making and continuous improvement.

1. Enterprise conversational AI and agentic automation platform (Electrolux)

• Built and maintained a large-scale, AI-powered conversational platform supporting end-to-end customer service journeys, including product support, service requests, and purchase interactions.
• Architected and deployed an intent-driven conversational assistant using Google Dialogflow CX, and later enhanced the platform with an AutoGen-based agentic AI framework to enable dynamic, context-aware, and autonomous resolution of complex customer queries.
• Led chatbot analytics, monitoring, and continuous optimization initiatives using BigQuery and Looker Studio, delivering executive-level KPI dashboards covering conversation volume, intent accuracy, fallback rates, and resolution performance.
• Designed and optimized prompt strategies for low-confidence and fallback scenarios, enabling graceful conversation recovery, accurate clarification flows, and improved resolution rates across both Dialogflow and agentic AI workflows.
• Implemented end-to-end PII detection and masking on customer conversation data using Google Cloud DLP, ensuring compliance with enterprise data privacy and security standards.
• Managed and optimized Databricks-based data pipelines for ingestion, transformation, and orchestration of conversational logs and operational analytics.
• Delivered measurable business impact by achieving 70% automation effectiveness across 10 European countries and 75% automation effectiveness across 2 North American countries.

2. Intelligent email classification and margin discovery platform (TESCO)

• Contributed to the development of an enterprise NLP solution to identify pricing discrepancies and margin recovery opportunities from buyer–supplier email negotiations across multiple geographies.
• Designed and implemented a multi-label email classification system to categorize promotional, contractual, and commercial communications, enabling early identification of revenue leakage.
• Built rule-based and dictionary-driven information extraction pipelines to capture critical financial attributes, including commercial value, fiscal period, and pricing terms.
• Analyzed continuous user feedback and model outputs to iteratively refine classification and extraction accuracy, improving overall system precision and recall.
• Led architectural enhancements to improve pipeline stability, scalability, and fault tolerance in a production enterprise environment.
• Delivered significant business impact by enabling the client to recover 60% more revenue compared to manual recovery processes.

3. MetaSearch Banking Platform (ANZ Bank)

• Designed and led the development of an enterprise-scale metadata management and intelligent search platform supporting core banking systems.
• Architected and deployed an internal AI-powered chatbot to streamline operational support by providing rapid resolutions for recurring system issues.
• Built and productionized NLP models to automatically extract problem statements, perform issue classification, and recommend actionable resolutions, improving support efficiency and response times.
• Collaborated with cross-regional stakeholders to align architecture, delivery, and operational requirements for Central Europe implementations.

4. Cognitive Tender Bidding Decision System(Dilip Buildcon Ltd, India)

• Designed a cognitive analytics solution to support bid/no-bid decision-making for large-scale infrastructure tenders across irrigation, road, and mining domains.
• Analyzed highly unstructured tender documents using IBM Watson Content Analytics and engineered custom NLP pipelines to extract key attributes such as sector classification, project cost, and bid submission deadlines.
• Developed a custom UIMA-based Java annotator to process and enrich annotated text, enabling data-driven tender evaluation and improving bidding accuracy and decision turnaround time.

6. Invoice Status Intelligence System(Dow Chemical, India)

• Built a cognitive document intelligence solution to identify invoice processing status from billions of unstructured textual records lacking consistent structure.
• Leveraged Watson Content Analytics to analyze free-form invoice comments and metadata, and developed a custom UIMA-based Java annotator to classify and extract document status indicators.
• Implemented robust parsing and annotation rules to enable timely invoice tracking and operational decision-making, significantly improving processing efficiency and visibility across invoice workflows.

Application Layer DDOS Attack Detection Using Hybrid Machine Learning Approach- International Journal of Security and its Application , vol 11, no 4, 2017, pp. 85-96. Journal indexed in Scopus

• Applied Machine Learning Course- Applied AI
• Agile methodology (IBM Training)
• Java Training (IBM Training)