Professional Experience
- Engineered a full-stack web app using React (component-based architecture) and FastAPI, optimizing backend performance via database indexing for 1K early users
- Built and optimized Python/SQL ETL pipelines for Vectorly's production platform using modular, AI-assisted code to automate data cleaning and transformation, improving compute efficiency by 60%
- Automated financial data workflows via Python/SQL ETL pipelines, improving reliability and cutting processing time by 60%, enabling faster user interactions
- Built a PDF-to-structured-data pipeline using Python (PyPDF, pandas) to standardize Excel inputs into 9+ unified schema fields; implemented RAG-based search for natural language queries over financial documents used by 30K+ UC Berkeley students
- Developed ML algorithms (k-means, decision trees, PCA, logistic regression) to perform unsupervised customer segmentation on 200,000+ users, increasing product penetration by 10%
- Built customer profiles of distinct user groups, enhancing targeted marketing and improving user engagement by 15%
- Engineered a geospatial ETL pipeline with RAG integration for natural language queries over LA County's COVID-19 workplace exposure dataset; combined vector embeddings with restaurant, employee count, and location metadata into a searchable model
- Implemented geofencing logic for restaurants within a 2-mile radius; served exposure statistics via API-driven UI components
Projects
MCP Server Tester
Architected an end-to-end web application enabling developers to register MCP servers, discover available tools via JSON-RPC protocol, and run configurable latency experiments. Calculated and visualized performance metrics including min/max latency, percentiles (p50, p95), and success rates.
DoorDash Wrap
Built feature pipelines modeling spend behavior and ordering patterns; applied clustering to surface personalized restaurant recommendations and year-end insights.
Maximus AI Hedge Fund
AI-native hedge fund system composed of autonomous agents using LLMs for trading decisions, combining financial analysis with cutting-edge AI technology.
Spam & Ham Classifier
Machine learning classifier using logistic regression to detect spam emails. Extracted features from over 8,000 emails and achieved 87% accuracy.
Housing Price Predictor
Linear regression model predicting home prices in Cook County. Addressed fairness and ethical taxation issues through statistical analysis.
Black Jack Simulator
Created a simulator to test the theory of Genetic Algorithm and practicing basic card counting
LSTM-Based Text Generation Model
Developed a character-level LSTM language model for next-character prediction, implementing a full training pipeline with sequence generation, hyperparameter tuning, and train/validation monitoring.
Yelp NoSQL vs SQL Data Processing Project
Analyzed and transformed semi-structured Yelp and college athletics datasets across MongoDB, Postgres, and Pandas, focusing on document modeling, performance optimization, and large-scale joins. Benchmarked NoSQL vs SQL join performance, showing Postgres outperforming MongoDB by two orders of magnitude.
Commodity Forecasting Challenge
Developed time-series forecasting models (XGBoost, LightGBM, LSTM) on multi-market data from LME, JPX, US Stocks, and Forex. Achieved top 15% cross-validation score on Kaggle competition leaderboard.
Predictive Housing Analytics in Cook County
Conducted data preprocessing and exploratory analysis on 200K+ housing records. Designed feature engineering pipelines with one-hot encoding and derived variables to optimize linear regression modeling.
Legal Document Intelligence Pipeline
Designed an end-to-end data pipeline that ingests unstructured legal documents (PDFs, contracts, pleadings), performs OCR, extracts key metadata, and loads structured results into Postgres + Elasticsearch. Built automated summarization, document classification, and deadline-flagging features using NLP, reducing manual review time by 60%.
Skills
Languages
Python, SQL, Java, JavaScript, HTML, CSS
Libraries
pandas, NumPy, scikit-learn, XGBoost, LightGBM, TensorFlow, Matplotlib, NLTK, spaCy
Data Engineering
ETL Pipelines, RAG, Vector Embeddings, MongoDB, PostgreSQL, Data Modeling
Web & APIs
React, FastAPI, Flask, REST APIs, JSON-RPC