Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

Hi, I’m Varsha Hemakumar, an AI Engineer and Machine Learning Scientist with 3+ years of experience designing, deploying, and scaling enterprise AI, Generative AI, and predictive analytics across fintech and large-scale enterprise environments. I specialize in LLMs, Retrieval-Augmented Generation (RAG), deep learning, and end-to-end MLOps, delivering business impact with over $10M in value, boosting model performance by 30–45%, and processing 10M–100M+ records in production. At Finta, Goldman Sachs, and KPIT Technologies, I’ve built production-grade AI platforms, scalable agent frameworks, and governance-driven ML lifecycles. I design cross-functional AI copilot solutions, implement model monitoring and compliance, and mentor teams to accelerate enterprise AI adoption while maintaining reliability and cost efficiency.…Hi, I’m Varsha Hemakumar, an AI Engineer and Machine Learning Scientist with 3+ years of experience designing, deploying, and scaling enterprise AI, Generative AI, and predictive analytics across fintech and large-scale enterprise environments. I specialize in LLMs, Retrieval-Augmented Generation (RAG), deep learning, and end-to-end MLOps, delivering business impact with over $10M in value, boosting model performance by 30–45%, and processing 10M–100M+ records in production. At Finta, Goldman Sachs, and KPIT Technologies, I’ve built production-grade AI platforms, scalable agent frameworks, and governance-driven ML lifecycles. I design cross-functional AI copilot solutions, implement model monitoring and compliance, and mentor teams to accelerate enterprise AI adoption while maintaining reliability and cost efficiency.

Varsha Hemakumar





Hi, I’m Varsha Hemakumar, an AI Engineer and Machine Learning Scientist with 3+ years of experience designing, deploying, and scaling enterprise AI, Generative AI, and predictive analytics across fintech and large-scale enterprise environments. I specialize in LLMs, Retrieval-Augmented Generation (RAG), deep learning, and end-to-end MLOps, delivering business impact with over $10M in value, boosting model performance by 30–45%, and processing 10M–100M+ records in production. At Finta, Goldman Sachs, and KPIT Technologies, I’ve built production-grade AI platforms, scalable agent frameworks, and governance-driven ML lifecycles. I design cross-functional AI copilot solutions, implement model monitoring and compliance, and mentor teams to accelerate enterprise AI adoption while maintaining reliability and cost efficiency.…Hi, I’m Varsha Hemakumar, an AI Engineer and Machine Learning Scientist with 3+ years of experience designing, deploying, and scaling enterprise AI, Generative AI, and predictive analytics across fintech and large-scale enterprise environments. I specialize in LLMs, Retrieval-Augmented Generation (RAG), deep learning, and end-to-end MLOps, delivering business impact with over $10M in value, boosting model performance by 30–45%, and processing 10M–100M+ records in production. At Finta, Goldman Sachs, and KPIT Technologies, I’ve built production-grade AI platforms, scalable agent frameworks, and governance-driven ML lifecycles. I design cross-functional AI copilot solutions, implement model monitoring and compliance, and mentor teams to accelerate enterprise AI adoption while maintaining reliability and cost efficiency.

Available to hire

Hi, I’m Varsha Hemakumar, an AI Engineer and Machine Learning Scientist with 3+ years of experience designing, deploying, and scaling enterprise AI, Generative AI, and predictive analytics across fintech and large-scale enterprise environments. I specialize in LLMs, Retrieval-Augmented Generation (RAG), deep learning, and end-to-end MLOps, delivering business impact with over $10M in value, boosting model performance by 30–45%, and processing 10M–100M+ records in production.

At Finta, Goldman Sachs, and KPIT Technologies, I’ve built production-grade AI platforms, scalable agent frameworks, and governance-driven ML lifecycles. I design cross-functional AI copilot solutions, implement model monitoring and compliance, and mentor teams to accelerate enterprise AI adoption while maintaining reliability and cost efficiency.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Intermediate

Work Experience

AI Engineer at Finta

September 1, 2025 - Present

Architected and led the end-to-end development of Aurora, an enterprise Generative AI copilot platform for deal intelligence, CRM automation, and relationship workflows using LLM orchestration, Retrieval-Augmented Generation (RAG), and contextual reasoning across millions of investor and contact records, improving workflow efficiency by 45%, reducing manual research by 60%, and accelerating revenue insights across $500M+ pipeline datasets. Designed and deployed a production-scale multi-model AI architecture on Google Vertex AI integrating GPT-class, Claude, and Gemini models with dynamic routing, fallback orchestration, and cost-latency optimization, improving response reliability by 35%, reducing inference costs by 28%, and enabling highly available enterprise AI services at scale. Built scalable AI agent infrastructure and MCP framework enabling cross-platform deployment across enterprise systems and developer environments, increasing AI adoption by 40%, enabling reusable AI workflow

AI & Data Science Engineer at Goldman Sachs

May 1, 2025 - August 31, 2025

Architected and productionized enterprise Generative AI and NLP pipelines using GPT-4, BERT, and Transformer architectures to process 50K+ regulatory and financial documents daily, improving information extraction accuracy by 36%, reducing analyst workload by 60%, and accelerating compliance intelligence across multi-million-dollar portfolios. Developed and deployed ML-based credit risk and fraud detection systems leveraging XGBoost, deep neural networks, and advanced feature engineering to monitor 12M+ weekly transactions, reducing false positives by 31%, improving classification precision by 28%, and preventing approximately $6.2M in annual fraud losses. Designed a scalable Retrieval-Augmented Generation (RAG) regulatory intelligence platform integrating vector databases, semantic search, and LLM orchestration to index and analyze 8TB+ structured and unstructured data, reducing research and audit preparation time by 70% and improving regulatory response SLAs by 55%. Implemented end-t

Machine Learning Engineer at KPIT Technologies

May 1, 2022 - August 31, 2024

Designed and deployed production-grade credit risk and fraud detection models using XGBoost, LSTM, and ensemble learning across 10M+ loan and transaction records, improving approval precision by 28%, reducing default exposure by $4M annually, and enabling real-time identification of 1,200+ fraud patterns per month. Developed advanced time-series forecasting and statistical modeling solutions for treasury and liquidity optimization across $50M+ working capital portfolios, improving forecast accuracy by 35% and supporting data-driven financial planning and executive decision-making. Built enterprise-scale NLP and Generative AI platforms leveraging GPT-based models, sentiment analysis, and Retrieval-Augmented Generation (RAG) pipelines to process 600K+ financial and customer documents annually, reducing manual analysis effort by 70% and accelerating reporting cycles from hours to minutes. Engineered computer vision and deep learning automation systems using OpenCV and neural architectures