Open to opportunities

Richa User

@richauser3

Message

Data Scientist focused on GenAI, causal inference, and MLOps—turning complex data into scalable decision intelligence.

United States

Message

What I'm looking for

I’m looking for a role where I can ship GenAI and ML systems in production—combining rigorous evaluation, reproducible data pipelines, and practical analytics—to improve real decision-making with measurable impact.

I’m a Data Scientist building practical, high-impact intelligence systems across GenAI, statistical modeling, and production data pipelines. My work blends dense retrieval, knowledge graphs, and LLM workflows with rigorous evaluation so models stay reliable as they scale.

Recently at CUNY Institute for State and Local Governance, I reduced legal charge standardization time from weeks to minutes across 16 U.S. jurisdictions by architecting a hybrid AI reasoning platform using Sentence-BERT embeddings, DSPy-optimized prompting, and multi-agent LLM workflows. I also increased automated classification agreement to 95%+ through ensemble decision systems with confidence calibration and continuous evaluation pipelines.

Earlier, at the NYC Office of the Mayor, I improved MWBE compliance risk detection by 32% while cutting manual audit effort by 45% using predictive ML across Python, SQL, and Power BI. I built an RAG platform that reduced procurement policy lookup time by 78% and accelerated AI-ready data preparation using modular ETL workflows.

Before that, as a Data Scientist at BYJU’s, I led causal inference and uplift modeling efforts—driving a 17% lift in student conversions and a 39% reduction in customer acquisition costs—while mentoring with strong analytical rigor as a Graduate Teaching Assistant. I care deeply about reproducibility, validation, and responsible deployment of AI systems.

Experience

Work history, roles, and key accomplishments

Current

Data Scientist

Current

CUNY Institute for State and Local Governance

May 2026 - Present (3 months)

Architecting a hybrid AI reasoning platform that reduces legal charge standardization time from weeks to minutes and improves automated classification agreement to 95%+ across multiple U.S. jurisdictions. Building reproducible nationwide demographic intelligence pipelines spanning 2014–2024 ACS data with automated validation and lineage tracking.

Hybrid AI Reasoning Knowledge Graphs DSPy Gradient Boosting Confidence Calibration Continuous Evaluation DuckDB Polars DBT Delta Lake Apache Arrow Python Data Validation Dagster Great Expectations Feature Engineering Lineage Tracking

Data Analyst

NYC Office of the Mayor

Oct 2025 - May 2026 (7 months)

Developed predictive ML models to improve MWBE compliance risk detection by 32% and reduce manual audit effort by 45% across city agencies. Built an RAG platform for natural-language procurement policy lookup and created anomaly detection and forecasting pipelines to improve utilization forecasting accuracy by 24%.

predictive modeling Python scikit learn XGBoost Pandas SQL Power BI Vector Search PostgreSQL REST APIs Feature Engineering Anomaly Detection Forecasting Prophet

Graduate Teaching Assistant

Baruch College, CUNY

Sep 2025 - May 2026 (8 months)

Supported analytics instruction for 80+ undergraduate and graduate students across Service Operations Management and Lean Six Sigma topics. Provided hands-on guidance in forecasting, hypothesis testing, regression, statistical process control (SPC), process capability analysis, DMAIC, and related statistical modeling methods.

Forecasting Hypothesis Testing Regression Modeling Statistics Process Control (SPC)Minitab Process Mapping Root Cause Analysis Inventory Optimization Quality Improvement Statistical Modeling Six Sigma Green Belt Methodologies Lean Six Sigma

Software Engineer

Gentrainer, MeyersWorkforce Solutions, LLC

Jun 2025 - Aug 2025 (2 months)

Built an XGBoost-based quality scoring model to reduce noisy AI training labels by 31% while improving prediction precision by 23%. Engineered MLflow-based model governance pipelines with explainability, calibration, drift detection, and experiment tracking, extending the system to benchmark LLM agent workflows using LangChain and DSPy.

XGBoost Python SQL scikit learn Optuna Feature Engineering Behavioral and Linguistic Features MLFlow SHAP Probability Calibration Performance Evaluation Lang Chain DSPy Orchestration

Data Scientist

Think and Learn Private Limited (BYJU's)

Sep 2018 - Nov 2024 (6 years 2 months)

Designed a Double Machine Learning causal inference framework to increase incremental student conversions by 17% by estimating individualized treatment effects and optimizing counseling interventions. Developed uplift modeling, policy simulation, and decision intelligence approaches to reduce customer acquisition costs by 39% and improve intervention effectiveness.