Vincent Kr

Experienced NLP Data Scientist with a passion for leveraging cutting-edge technologies to solve complex business problems, I possess a strong background in machine learning, natural language processing, and text analytics. With over 10 years of experience in the field, I have a deep understanding of state-of-the-art NLP techniques. I have experience in designing and implementing end-to-end NLP pipelines, from data collection and cleaning to model deployment and performance monitoring.

My strong analytical and problem-solving skills, combined with my ability to work collaboratively in cross-functional teams, make me a valuable asset to any organization seeking to leverage the power of NLP to drive business growth and customer satisfaction.

Senior Data Scientist - Mudano part of Accenture - London, UK

(2022-01)

• Senior Data Scientist: Content Generation Service for Regulatory Drug Submissions: Developed an AI-powered content generation platform on Azure to automate the creation of regulatory submission documents for drug approval. The platform uses multi-agent systems to generate content by mapping research and drug documentation to standard regulatory templates with instructions.
• AI-Driven Content Generation: Built an automated content generation system using AutoGen on Azure, leveraging the GPT-4o model with an actor/critic approach—one agent generates content while another reviews and provides feedback. This system automates the creation of first draft submission documents, accelerating the drug approval process.
• Platform Architecture & Deployment: The system was deployed using Azure services and Databricks for orchestrating Python-based AI pipelines exposed as APIs.
• Document Parsing & Content Extraction: Developed pipelines to extract and organize content—including text, figures, and tables—from multiple document formats using Azure Document Intelligence. Implemented semantic chunking with Azure AI Search using the HNSW algorithm for efficient retrieval.
• MLOps Pipelines: Created MLOps pipelines to record agent input/output pairs, track latency, and monitor metrics like compliance adherence and content accuracy, enabling systematic performance evaluation.
• Tech Stack: Python, FastAPI, Databricks (pipeline orchestration & API deployment), Azure, Azure AI Search, Azure Document Intelligence, Azure OpenAI, MLFlow, Autogen, Azure SQL Database, Langchain, Prompt engineering
• Lead Data Scientist: Intelligent Email Classification/Drafting: The project aimed to utilize AI for intelligently routing and classifying emails based on content, while also drafting responses by validating trade statuses
• Lead a team of Data Scientists and trained a large language model to intelligently draft (Mistral-7b) emails about post-trade processing queries for a leading UK bank, resulting in a 70% reduction in manual classification/email routing and drafting by the operations team with 90% accuracy
• Designed, developed and productionized a custom trained large language model (BERT) to intelligently route emails about post-trade processing queries along with market/region extraction resulting in a 95% reduction in manual classification/email routing by the ops team.
• Built an MLops pipeline to capture misclassified emails, leading to timely retraining of the classification model for improved accuracy and performance.
• Technical Skills: Python, Pandas, SQL, Sagemaker, AWS, Lambda, ECR, Llama-7b, Mistral-7b, BERT, Pytorch, AWS Jumpstart
• Senior Data Scientist: Semantic AI - Supply Chain Shortages: The focus of the project was to create and manage supply chain shortages data into a knowledge graph, leveraging Large Language Models (LLM) to dynamically answer questions and generate knowledge graphs based on user inquiries.
• Developed a semantic layer to accurately interpret and process user questions, ensuring precise and relevant responses.
• Knowledge Graph Generation: Constructed a Knowledge Graph (KG) that dynamically updates based on user questions, highlighting gaps in data and improving the system's ability to provide accurate information on gaps in supply chain.
• Technical Skills: Python, neo4j, Snowflake, Snowflake Cortex, FastAPI

Data Scientist - Aqovia/NeuerEnergy - London, UK

(2019-06 - 2021-12)

• Implemented a short term (7 days ahead) wholesale electricity price forecast algorithm for UK, France, Spain and Danish markets with high accuracy.
• Developed and implemented Python-based statistical models to quantify volume risk in Power Purchase Agreements (PPAs) for renewable energy projects, incorporating historical production data, weather patterns, and contractual terms to enhance risk management strategies.
• Developed predictive maintenance algorithm to predict the failure of a Grid battery storage system 24 hours earlier which results in the reduction of 27% of downtime.
• Implemented a prediction model to accurately classify the type of alarms and the expected downtime of the Grid storage system using SCADA data and the alarm log files.
• Worked on collecting, cleaning, analyzing and modeling of large power generation, energy consumption, maintenance activities and all the machine parameters datasets with high variability programming language like SQL and Python
• Defined metrics and built monitoring dashboards using PowerBI for continuous management of model performance.
• Technical Skills: Python, Pandas, Azure ML Service, PowerBI

Data Migration Specialist - TCS - Chennai, India

(2012-08 - 2018-07)

• Successfully migrated a Telecom's Collection system from Salesforce to PEGA with 3 million customers involving complex business mappings of data in PEGA database using SQL.
• Used ETL methodologies and best practices to perform Data Migrations, Extraction, Transformation and Loading using Netezza and IDQ.
• Developed alert scripts to monitor failures in Database load, changes in data types from multiple data sources.
• Audited, improved and cleaned the existing legacy Debt Collection agency (DCA) data to adapt into the new application database, helping companies from getting cleaned data from DCA third parties.
• Technical Skills: SQL, Netezza, IDQ, Shell Script

MSc - Data Science - University of Essex (2018-09 - 2019-09)

BTech - Mechanical Engineering - Pondicherry Engineering College (2008-08 - 2012-05)