Anil Kumar C
Summary
GenAI Engineer and AI/ML Developer with a specialization in designing and deploying advanced LLM-based solutions, enterprise chatbots, and Retrieval-Augmented Generation (RAG) pipelines. Proven success in implementing scalable AI architectures and fine-tuned models for secure, intelligent automation. Built and deployed enterprise-grade conversational AI systems using OpenAI GPT-3.5/4, FinBERT, and HuggingFace Transformers for tasks like semantic search, summarization, and contextual Q&A. Architected robust RAG pipelines using LangChain, FAISS, and Pinecone for domain-specific document retrieval and real-time AI responses. Expertise in prompt engineering, context handling, and chain of thought design to ensure accurate, grounded outputs in production-grade LLM applications. Delivered full-stack chatbot platforms with secure frontends via Streamlit and Flask, supporting enterprise users across medical, finance, and legal domains. Integrated AI workflows with Azure OpenAI, Azure Cognitive Services, Cosmos DB, Azure AD, Azure Search, and Azure SQL to ensure performance, scalability, and compliance. Contributed to open-source inspired internal tools for reusable LangChain chains, prompt libraries, and embedding strategies. Collaborated with cross-functional teams to align AI models with domain-specific needs and regulatory standards. Designed secure ingestion workflows with access control and metadata tagging to enable responsible AI access. Led workshops, architecture reviews, and proof-of-concept demonstrations to scale GenAI usage across enterprise functions.
Overview
15
15
years of professional experience
Work History
GenAI/ML Engineer
Regeneron Pharmaceuticals, Inc.
07.2024 - Current
- Project: RAG-based AI Assistant for Pharma Knowledge Retrieval
- Objective: Designed a secure, scalable Retrieval-Augmented Generation (RAG) system for internal medical and research teams to query large collections of unstructured documents (PDFs, SOPs) using natural language.
- Tools & Tech: LangChain, OpenAI GPT-4, FAISS, Pinecone, Python, Streamlit, HuggingFace Transformers, Azure OpenAI, Azure Search, Azure SQL, Azure Cognitive Services
- Responsibilities:
- Developed production-grade LangChain pipelines integrating OpenAI GPT models with FAISS/Pinecone for RAG-based semantic search over enterprise content.
- Engineered prompt templates with LangChain PromptTemplate and ConversationMemory to support multi-turn, context-aware chat interactions.
- Applied recursive and sentence-based chunking strategies with token control to improve embedding precision and document segmentation.
- Delivered a secure Streamlit interface for users to upload documents, ask questions, and receive grounded, cited LLM responses.
- Integrated Azure OpenAI, Cognitive Services, Azure Search, and Cosmos DB to ensure performance, compliance, and scalability.
- Designed ingestion pipelines supporting multi-format content with metadata tagging and access governance.
- Benchmarked performance (latency, precision@k, hallucination rate) and refined prompts to optimize LLM output quality.
- Integrated Azure Cognitive Search to enhance retrieval accuracy and relevance ranking for pharma-specific terminologies and multilingual documents.
- Designed and executed unit tests and regression tests for LLM-driven prompts and retriever pipelines, ensuring stability and reproducibility across different data loads.
- Collaborated with data scientists and SMEs to align model behavior with compliance and pharma-specific knowledge constraints.
- Documented architecture, retriever logic, embedding workflows, and prompt libraries for future reuse and enterprise onboarding.
- Project: RAG-based AI Assistant for Pharma Knowledge Retrieval
- Objective: Designed a secure, scalable Retrieval-Augmented Generation (RAG) system for internal medical and research teams to query large collections of unstructured documents (PDFs, SOPs) using natural language.
- Tools & Tech: LangChain, OpenAI GPT-4, FAISS, Pinecone, Python, Streamlit, HuggingFace Transformers, Azure OpenAI, Azure Search, Azure SQL, Azure Cognitive Services
GenAI/ML Engineer
USAA (United Services Automobile Association)
11.2023 - 07.2024
- Project: Financial Sentiment Analyzer using FinBERT
- Objective: Developed a machine learning–based financial sentiment analysis system using FinBERT to help analysts and stakeholders assess sentiment in news headlines and market data. The solution enabled better investment decisions and risk evaluation by classifying market signals into positive, neutral, or negative sentiment with contextual accuracy.
- Tools & Tech: Python, FinBERT, HuggingFace Transformers, LangChain, FAISS, Streamlit, AWS EC2, Pandas, SQL
- Responsibilities:
- Developed RAG pipelines using LangChain and FAISS for intelligent question-answering over financial disclosures and market reports.
- Configured FAISS vector store with optimized embeddings and chunking logic using HuggingFace Transformers to enable low-latency semantic retrieval.
- Implemented parsing and sentence-based chunking strategies to improve embedding granularity and retrieval accuracy.
- Created prompt templates with dynamic context injection to support follow-up intent, sentiment explanation, and reasoning chains.
- Built a secure Streamlit interface with document upload, sentiment classification, and explainable AI outputs, tailored to financial analysts.
- Built ingestion pipelines to handle structured/unstructured files with access controls and document tagging.
- Benchmarked API latency, token usage, hallucination mitigation, and precision@k for ongoing LLM prompt optimization.
- Conducted continuous evaluation with SMEs to validate financial sentiment accuracy and refine classification strategies.
- Documented full architecture, prompt libraries, embedding strategies, and system design; delivered KT sessions to downstream teams.
- Integrated FinBERT outputs with downstream analytics workflows using Pandas for further investment risk analysis.
- Collaborated with product teams to extend the system for real-time ingestion of RSS feeds and market data APIs.
- Project: Financial Sentiment Analyzer using FinBERT
- Objective: Developed a machine learning–based financial sentiment analysis system using FinBERT to help analysts and stakeholders assess sentiment in news headlines and market data. The solution enabled better investment decisions and risk evaluation by classifying market signals into positive, neutral, or negative sentiment with contextual accuracy.
- Tools & Tech: Python, FinBERT, HuggingFace Transformers, LangChain, FAISS, Streamlit, AWS EC2, Pandas, SQL
ML Engineer – Demand Forecasting
Juniper Networks, Sunnyvale, CA
03.2020 - 09.2023
- Project: ML-Driven Shortage & Excess Analysis System
- Objective: Developed a machine learning solution to predict raw material demand, identify shortages, and flag excess inventory. The system helped supply chain and procurement teams proactively place orders or reallocate stock to global markets to optimize costs and fulfillment timelines.
- Tools & Tech: Python, Scikit-learn, Pandas, NumPy, SQL, Oracle DB, Matplotlib, Streamlit
- Responsibilities:
- Collected and preprocessed historical procurement and consumption data using Pandas and SQL from Oracle DB, cleaning and transforming it for ML modeling.
- Built predictive models (Random Forest, XGBoost) to forecast material demand based on historical trends, lead times, and product lifecycles.
- Engineered features for seasonality, supplier risk, and product velocity to improve forecast accuracy and shortage detection.
- Designed logic to flag excess inventory and suggest redistribution opportunities across regional warehouses using statistical thresholds.
- Developed a Streamlit dashboard for business teams to visualize demand predictions, risk scores, and material recommendations.
- Tuned models using GridSearchCV, improving prediction accuracy and reducing stockout risks in critical material categories.
- Created scheduled Python ETL scripts to refresh predictions daily and log model performance for ongoing monitoring.
- Partnered with supply chain managers and data stakeholders to validate outputs and integrate insights into order planning workflows.
- Project: ML-Driven Shortage & Excess Analysis System
- Objective: Developed a machine learning solution to predict raw material demand, identify shortages, and flag excess inventory. The system helped supply chain and procurement teams proactively place orders or reallocate stock to global markets to optimize costs and fulfillment timelines.
- Tools & Tech: Python, Scikit-learn, Pandas, NumPy, SQL, Oracle DB, Matplotlib, Streamlit
Tableau Developer
Caterpillar, Bangalore India
08.2016 - 02.2020
- Project: Container Cross Dock Operations, OHIO
- Tools Used: Tableau Developer and Oracle DB
- Responsibilities:
- Involved in requirement analysis, data analysis, and business analysis of existing systems and utilized Tableau's capabilities, such as data extracts, data blending, forecasting, dashboard actions, and table calculations.
- Created optimized technical design documents for specific requirements.
- Performed data analysis using SQL and designed data models.
- Developed report-level logic as per the requirement and database constraints, creating proofs of concept (POCs) for client approvals.
- Created customized interactive dashboards in Tableau using marks, actions, filters, parameters, and calculations.
- Developed dashboards with context/global filters, calculated fields, and parameters that allowed for tracking and improving business data for Packaging and Transportation using Tableau.
- Improved operational efficiency by 30% through the analysis of data performance, data quality, error rates, and root cause analysis.
- Implemented versioning of reports based on client feedback and incorporated improvements.
- Handled performance tuning of reports to reduce report rendering time.
- Managed change requests and provided support and maintenance for deployed projects.
- Project: Container Cross Dock Operations, OHIO
- Tools Used: Tableau Developer and Oracle DB
Tableau Developer
Robert Bosch Engineering and Business Solutions
08.2015 - 07.2016
- Tools Used:
- Tableau Desktop and SQL Server
- Responsibilities:
- Interfaced with the business folks to understand the data and business requirements by connecting with them.
- Converted screens into documentation for reference and prepared data for data visualization.
- Connected data to data sources using Tableau Desktop. Created data visualizations by exploring Dimensions, Measures, and Calculations to achieve better charts that matched the requirements.
- Developed workbooks and hyper extracts on Tableau Desktop.
- Published workbooks and data sources to Tableau Development and QA sites.
- Refreshed extracts on Development and QA sites. Monitored the status of extract schedules on the QA site. Performed unit testing on Development and QA sites.
- Resolved any data or performance issues related to workbooks and data sources.
- Moved to the Production server to bring it live to the users.
- Presented to larger user groups once the data visualization was ready with the needed insights as per the expectation.
- Monitored Usage Reports of their Workbooks and Data Sources.
- Worked on continuous improvements as per the business folks to meet their business needs.
- Monitor data source performance and troubleshoot any issues.
- Tools Used:
- Tableau Desktop and SQL Server
Informatica/PLSQL Developer
Wipro Technologies Pvt. Ltd., United States
04.2011 - 06.2015
- Project – Best Buy DWH, US
- Tools Used:
- Informatica Power Centre and SQL / Flat files
- Responsibilities:
- Analyzed data in source systems and analyzed targets, transformed, mapped data, and loaded data into targets using Informatica Power Center in accordance with TDDs and specifications.
- Designed project-related documents like Low-Level Design (LLD).
- Extensively wrote user-defined SQL code for overriding generated SQL queries.
- Actively participated in team meetings to gather business requirements and in developing specifications.
- Involved in unit testing and system testing.
- Participated in modifying initial loading mappings in INFORMATICA POWERCENTER 8.6.0 and contributed to the development of incremental load mappings.
- Developed workflows and worklets for respective mappings. Prepared the respective unit test cases and executed test cases for mappings.
- Project – Best Buy DWH, US
- Tools Used:
- Informatica Power Centre and SQL / Flat files
Education
Bachelor of Technology - Electronics and Communications Engineering
Jawaharlal Nehru Technological University
Anantapur, AP05.2010
Skills
- Programming & Frameworks: Python, LangChain, OpenAI API, HuggingFace Transformers, Flask, Streamlit
- NLP & LLMs: BERT, FinBERT, RoBERTa, Prompt Engineering, Sentence Transformers
- Vector Stores & Retrieval: FAISS, Pinecone, Vector Embeddings, RAG Architecture
- Cloud & GenAI Services: Azure OpenAI, Azure Cognitive Services, Azure AD, Azure Search, Azure SQL, Cosmos DB
- Databases: Snowflake, Oracle and SQL Server
- ETL & BI Tools: Alteryx, Tableau and Power BI
Timeline
GenAI/ML Engineer
Regeneron Pharmaceuticals, Inc.
07.2024 - Current
GenAI/ML Engineer
USAA (United Services Automobile Association)
11.2023 - 07.2024
ML Engineer – Demand Forecasting
Juniper Networks, Sunnyvale, CA
03.2020 - 09.2023
Tableau Developer
Caterpillar, Bangalore India
08.2016 - 02.2020
Tableau Developer
Robert Bosch Engineering and Business Solutions
08.2015 - 07.2016
Informatica/PLSQL Developer
Wipro Technologies Pvt. Ltd., United States
04.2011 - 06.2015
Bachelor of Technology - Electronics and Communications Engineering
Jawaharlal Nehru Technological University