Kalyani Vinayagam Sivasundari
Ai Software Engineer @KGS Technology Group, Inc
Signup · Get unlimited contacts
WORK HISTORY
Ai Software Engineer @KGS Technology Group, Inc
Alpharetta, GA, US
Developed microservice-based sales and inventory workflows using REST APIs, automating key modules in the Enterprise Management System. Built AI agents with OpenAI APIs for lead scoring and demand forecasting, and delivered a LangChain-powered Streamlit dashboard visualizing real-time KPIs and sales insights.
EDUCATION
Coimbatore Institute of Technology
Bachelor of Engineering - BE, Computer Science and Engineering
The George Washington University
Master of Science - MS, Computer Science
ABOUT KALYANI VINAYAGAM SIVASUNDARI
Software Engineer and Machine Learning Specialist skilled in building scalable, intelligent systems at the intersection of AI, cloud, and data engineering. Experienced in designing microservice architectures, RAG pipelines, and LLM-based automation using Python, AWS, and modern AI frameworks.With hands-on experience across KGS Technology Group, FactEntry, Frizb, and Samsung, I focus on transforming complex data into actionable intelligence — from enterprise automation to multimodal document analytics.Driven by innovation, I specialize in Generative AI, NLP, and intelligent automation, building systems that deliver measurable business impact. Expertise• Languages : Python, C, C++, SQL• Databases : MySQL, PostgreSQL• Libraries : Numpy, Pandas, Skicit-Learn, Keras, XGBoost• Frameworks : PyTorch, Tensorflow, HuggingFace Transformers• Visualization & Analytics : Power BI, Matplotlib, Seaborn, Plotly • Generative AI : BERT, RAGs, LangChain, Prompt Engineering • Cloud & DevOps : AWS, Docker, Kubernetes, GitHub Actions • Tools : Git, GitHub, UML, Postman, VSCode, Streamlit, ANTLR PublicationIntegrating OpenAI Whisper with AASIST for DeepFake Audio Detection — Presented at HCII 2024 Conference, achieving EER of 8.67% on ASVspoof 2021 DF subset and 20.91% on “In-the-Wild” datasets, outperforming prior benchmarks Accomplishments AI Multi-Agent Text2SQL System (Frizb): Achieved ~90% accuracy in dynamic SQL query generation using Bedrock embeddings and fine-tuned LLMs Financial Data Extraction (FactEntry): Improved data extraction relevance by 35% through advanced RAG-based retrieval🧠 Deep Learning Research : Built multimodal table extraction systems using LLaMA-4 and Table Transformer with 92% precision Conference Recognition : Published and presented deepfake detection research on large-scale neural architectures
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.