Open to opportunities

Sai Charan Tammineni

@saicharantammineni

Message

I build production GenAI, NLP, and machine-learning systems that reduce costs and improve accuracy.

United States

Message

What I'm looking for

I'm looking to build production GenAI, NLP, and machine-learning systems where I can improve model accuracy, reduce operational costs, and automate high-impact enterprise workflows.

I'm building Azure-hosted NLP automation at Socratics AI that classifies 10,000+ unstructured financial-statement line items into standardized accounting taxonomies with 96% held-out accuracy.

I've reduced LLM inference costs by 75% through RAG, embeddings, vector databases, semantic search, and selective Anthropic API routing. I also cut investment-banking financial-model generation from 8–12 hours to under 30 minutes and reduced manual QA checks from half a day to about two hours with reusable Claude Code agents.

At Verint Systems, I integrated 8M+ call, chatbot, and agent-assist records, improved campaign ROI forecasting MAPE by 22%, and identified onboarding changes that increased 90-day feature adoption by 18%.

Earlier at CGI, I built fraud classifiers and anomaly-detection workflows supporting 200K+ pension beneficiaries, deployed automated batch scoring to Azure, and reduced SQL query execution from 11 seconds to under four seconds across 15 reporting workflows.

Experience

Work history, roles, and key accomplishments

Current

Data Scientist

Current

Socratics AI

Aug 2025 - Present (1 year)

Engineered and deployed an Azure-hosted NLP pipeline achieving 96% accuracy on financial-statement classification, reducing LLM inference costs by 75% and accelerating model generation from 8-12 hours to under 30 minutes. Fine-tuned BERT and FinBERT with LoRA/QLoRA, built Claude Code agents for validation, and used AgentViews for failure-mode analysis.

Python PyTorch BERT FinBERT LoRAs QLoRA RAG Vector Databases Pinecone LangGraph Langsmith Claude Code Azure Docker Kubernetes NLP LLM Evaluation Hallucination Detection

Data Scientist

Verint Systems

May 2024 - Aug 2025 (1 year 3 months)

Integrated call and chatbot logs across 8M+ records to measure performance, reduced campaign ROI forecasting MAPE by 22%, and productionized forecasting pipelines with Docker and GitHub Actions. Segmented user sessions and analyzed A/B tests to increase feature adoption by 18%.

Python SQL Pandas NumPy scikit learn regression Feature Engineering Docker GitHub Actions Azure A B Testing Hypothesis Testing Clustering CI CD

Data Scientist

CGI

Mar 2020 - Jul 2023 (3 years 4 months)

Developed and evaluated fraud classifiers using Logistic Regression, XGBoost, and LightGBM, designed A/B tests for fraud-alert thresholds, and implemented autoencoder-based anomaly detection. Deployed models to Azure VMs via GitHub Actions and optimized SQL data models with Kimball star schemas.