• Designed and developed an algorithmic trading framework for intraday strategies using Python and Alpaca API, focusing on sideways trend detection and real-time risk management.• Built backtesting capabilities to simulate and optimize strategy performance, incorporating stop-loss mechanisms and dynamic position sizing.• Implemented a data collection system to track slippage during live trading, enabling analysis of execution performance and strategy improvement.• Streamlined high-frequency data processing with Pandas and NumPy, enabling efficient handling of level 1 trade and quote data for precise execution.• Leveraged AI-driven development tools for enhanced troubleshooting, code iteration, and strategy refinement, utilizing ChatGPT and Claude interfaces.

• Created a workflow for Named Entity Recognition (NER) dataset annotation of patient lab result comments. Implemented fine-tuning of Spark NLP model pipelines. • Optimized data processing with PySpark. Increased confidence scores of Hugging Face NER models by ~3% through tokenization improvements.• Debugged, tuned, and compared classification model variants for Alzheimer's Early Signal Identification as part of an ongoing collaboration with Eli Lilly and Co.

• Implemented a Proof-of-Concept distributed ETL and stream processing pipeline on 3 Linux servers using Confluent for Apache Kafka. • Collaborated with the Database team to ensure correctness of the ksqlDB data model for creating fast materialized views.• Set up multithreaded Kafka connectors to introduce real-time incremental refresh to business-critical reporting tables in Microsoft SQL Server and Cassandra, reducing data latency from ~40 min to less than 1 min.

• Constructed training datasets of ~400,000 distinct container shipments by joining data from 10+ tables stored in Microsoft SQL Server. • Identified impactful data fields and tables by organizing discussions with members of IT and Operations.• Reduced Mean Absolute Error by 60% over baseline models using ML time series regression for predicting the Time-of-Arrival of container shipments.

• Participated in a training-focused internship, acquiring hands-on experience in FICO scorecard development and credit risk analysis for financial applications.• Developed predictive models using Python, focusing on FICO scorecard techniques to assess the creditworthiness of loan applicants.• Collaborated remotely with student teams in China, enhancing communication skills and cross-cultural understanding within the context of the financial industry.

• First author of conference publication: "Chronic Respiratory Disease: Risk Modeling Potential and Limitations," IEEE BigData 2021 [ https://ieeexplore.ieee.org/document/9672074 ]• Combined spatiotemporal datasets of 10+ environmental and demographic variables into ~600K data points to predict monthly CRD mortality rates in the U.S.• Aggregated GIS raster datasets into ~3,000 U.S. counties using GeoJSON shapefiles and geospatial Python libraries while ensuring algorithmic efficiency.• Conducted iterative model evaluations and algorithmic feature selection to refine predictive accuracy, gaining practical experience in ML regression model optimization.• Facilitated regular meetings with advisor, ensuring progress on milestones and effective guidance on research directions.