Deekshit Atluri
Grand Rapids
Summary
Results-driven Data Scientist with a strong foundation in machine learning, predictive modeling, and data visualization. Proven ability to develop impactful models and analytical reports that enhance decision-making and operational efficiency.
Overview
5
5
years of professional experience
Work History
Data Scientist
Optum
Grand Rapids
09.2025 - Current
- Operationalized anomaly-detection models on enterprise device telemetry to identify abnormal sensor behavior and surface high-risk devices through weekly executive reporting.
- Performed feature-level analysis of sensor triggers, device utilization, and anomaly scores to prioritize potential operational incidents and reduce reactive escalations.
- Analyzed enterprise system-usage patterns, identifying underutilized assets and cost drivers to support infrastructure-spend optimization.
- Developed GenAI-based incident-classification pipeline using Azure Machine Learning and Databricks, standardizing incident reports for improved categorization.
- Grounded incident-classification outputs using historical reports, current incident data, and enterprise taxonomy to improve categorization consistency and downstream reporting.
- Designed A/B tests for prompt and model configurations, evaluating classification quality, response consistency, and failure modes across controlled test sets.
- Applied prompt engineering techniques including few-shot prompting, contextual grounding, structured-output schemas, and instruction tuning to improve model reliability.
- Built a natural-language-to-SQL workflow that translates business questions into Snowflake queries and returns contextual responses through an existing Copilot interface.
- Validated generated SQL for schema alignment, join logic, filters, aggregations, and result accuracy before enabling conversational access to enterprise data.
- Produced analytical reports on high-frequency sensor triggers and anomalous device behavior, informing business and technology leadership on enterprise risk patterns.
Data Analyst
Vivace pact tech
Grand Rapids
04.2025 - 09.2025
- Analyzed financial, transaction, customer, and loan datasets to support business reporting, risk analysis, and predictive modeling.
- Built Python and SQL workflows to clean, transform, reconcile, and validate data from internal databases, spreadsheets, and third-party sources.
- Developed classification and regression models using XGBoost, Random Forest, LightGBM, CatBoost, and Logistic Regression for customer risk, churn, fraud indicators, and loan-performance use cases.
- Performed feature engineering, normalization, categorical encoding, aggregation, and class balancing with SMOTE to prepare model-ready financial datasets.
- Evaluated models using precision, recall, F1-score, ROC-AUC, RMSE, and MAE and tracked experiments and model versions using Azure Machine Learning and MLflow.
- Developed Power BI and Tableau dashboards to visualize revenue, transaction trends, customer behavior, loan performance, and financial risk indicators for informed decision-making.
- Wrote advanced SQL queries using joins, CTEs, subqueries, window functions, and aggregations for recurring reports and ad hoc financial analysis.
- Implemented Python and SQL data-quality checks to detect duplicates, missing values, key mismatches, invalid mappings, and reconciliation differences.
- Performed customer segmentation, churn, and lifetime-value analysis to enhance customer targeting and retention strategies.
- Partnered with company leadership and business teams to translate financial inquiries into actionable analytical reports, dashboards, and model-based recommendations.
Deep Learning Research Assistant
Grand Valley state University
Grand Rapids
07.2024 - 12.2024
- Led a three-member medical computer-vision research team developing YOLO-based object-detection models for gastrointestinal polyp localization and Vision Transformer models for malaria classification from blood-smear images.
- Fine-tuned pretrained YOLO and ViT architectures in TensorFlow/Keras on domain-specific medical-image datasets, adapting model inputs and prediction layers for object detection and binary image classification.
- Optimized network architecture using Conv2D, Conv2DTranspose, batch normalization, embedding layers, ReLU, and sigmoid activations to improve feature extraction and classification behavior.
- Performed hyperparameter tuning across batch size, input resolution, learning rate, training epochs, and model dimensions, selecting configurations based on validation performance and prediction quality.
- Conducted error analysis on false positives, false negatives, missed detections, and misclassified samples to identify issues with class imbalance and low-contrast regions, informing model refinement.
- Evaluated neutrosophic image preprocessing as a technique to enhance low-contrast and ambiguous regions in medical images, contributing to improved model accuracy.
- Performed comparative benchmarking of YOLO and Vision Transformer configurations across image- and video-based datasets, using validation accuracy and detection outputs to guide model selection.
- Distributed deep-learning workloads across Google Colab GPUs, Google Cloud TPU v3 resources, and university-hosted GPU infrastructure to accelerate training and support larger experimental runs.
- Applied model checkpointing and iterative evaluation to preserve high-performing training states and compare predictions across multiple training cycles.
- Managed source code and experiment revisions through Git and Visual Studio Code, coordinated technical tasks for two research interns, and presented model findings and optimization recommendations to faculty advisor every three days, ensuring project alignment and progress.
Data Analyst Intern
PSIMS
Gannavaram
08.2021 - 07.2022
- Created and updated Power BI dashboards used by the finance team to review departmental expenses, revenue trends, budget variances, and monthly financial performance.
- Pulled financial data from SQL Server, Oracle, MySQL, and Excel files and combined the results into reporting datasets for Power BI and SSRS.
- Wrote SQL queries using joins, CTEs, subqueries, aggregate functions, CASE statements, and window functions to extract and summarize finance data by department, account, and reporting period.
- Used Power Query to clean source files, standardize column formats, merge data from multiple reports, and reduce the amount of manual preparation required each month.
- Created DAX measures in Power BI for budget-versus-actual variance, percentage change, monthly trends, running totals, and prior-period comparisons.
- Used Python with pandas and NumPy to clean financial datasets, remove duplicate records, handle missing values, validate account mappings, and compare totals across source systems.
- Reconciled financial reports by comparing Excel files, database extracts, and Power BI results, then investigated differences in transaction counts, account balances, and departmental totals.
- Built and maintained fact and dimension tables in SQL Server to organize transaction, department, account, and calendar data for reporting.
- Created parameterized SSRS reports for recurring finance requests, allowing users to filter results by department, date range, account category, and reporting period.
- Met with finance team members to understand reporting needs and converted those requirements into SQL queries, Power BI visuals, DAX calculations, and report filters.
- Tested dashboard calculations and report outputs before release by checking totals against source files and documenting any data issues found during validation.
- Tracked reporting defects and enhancement requests in JIRA and maintained report definitions, business rules, and data-mapping documents in Confluence and SharePoint.
- Used Git to maintain versions of SQL scripts, Python files, validation logic, and ETL documentation as reporting requirements changed.
Education
Master of Science - Data Science
Grand Valley State University
Allendale, MI12-2024
Skills
- Predictive modeling
- Machine learning
- Deep learning
- Anomaly detection
- Feature engineering
- Statistical analysis
- SQL query writing
- Data visualization
Timeline
Data Scientist
Optum
09.2025 - Current
Data Analyst
Vivace pact tech
04.2025 - 09.2025
Deep Learning Research Assistant
Grand Valley state University
07.2024 - 12.2024
Data Analyst Intern
PSIMS
08.2021 - 07.2022
Master of Science - Data Science
Grand Valley State University