Open to opportunities

Venkatesh Shamala

@venkateshshamala

Message

I build production AI, LLM, and fraud detection systems for high-scale financial transactions.

United States

Message

What I'm looking for

I'm looking to build production AI, LLM, NLP, and machine-learning systems that improve fraud detection, risk decisions, and enterprise automation at scale.

At Stripe, I build enterprise RAG, fraud detection, and real-time risk-scoring systems that support natural-language document querying and coordinated fraud-ring detection. My work processes 12M+ transactions monthly, reduced fraud leakage by 18%, and cut risk-scoring latency from 420ms to 180ms.

I develop secure Python/FastAPI services and hybrid retrieval systems using FAISS, Pinecone, PostgreSQL, GPT-4, BERT, LangChain, and fine-tuned Hugging Face models. I also build governed MLOps workflows with MLflow, Airflow, Docker, Kubernetes, Terraform, AWS, and GCP, maintaining 99.95% uptime while reducing concept drift by 28%.

Previously at Zensar Technologies, I built credit-risk, default, churn, forecasting, and anomaly-detection models for banking, improving loan prediction AUC from 0.81 to 0.91 for 2M+ applications.

Experience

Work history, roles, and key accomplishments

Current

Applied LLM Data Scientist

Current

Stripe

Nov 2024 - Present (1 year 9 months)

Designed and developed enterprise-grade GenAI and fraud detection systems combining RAG architecture, backend services, NLP, graph analysis, and real-time ML pipelines. Launched a GenAI chatbot with RAG architecture, reducing manual lookup time by 45% and improving response relevance.

Python SQL PyTorch TensorFlow XGBoost LightGBM CatBoost GPT 4 BERT Hugging Face Transformers Faiss Pinecone MLFlow SHAP W&B Terraform Docker Kubernetes Airflow AWS GCP PEFT LoRA Prompt Engineering Model Governance

Machine Learning Engineer

Zensar Technologies

Jan 2020 - Aug 2023 (3 years 7 months)

Built data science and analytics solutions for credit scoring, default risk prediction, churn modeling, and portfolio optimization in banking. Developed high-performing credit risk models increasing AUC from 0.81 to 0.91 for 2M+ loan applications.