Stuti Upadhyay

Results-driven AI Engineer and Data Scientist with an M.P.S. in Data Science, specializing in building production ML systems and multi-agent AI frameworks. Successfully enhanced enterprise capabilities by deploying innovative solutions in computer vision and NLP, while fostering collaboration to transform complex business challenges into effective data strategies.

Finance Based Agent, Personal Project | github.com/stutiupadhyay03/Finance_Based_Agent | CrewAI, Gemini 2.5 Pro, LiteLLM, yfinance, Python

• Built a 7-agent autonomous financial intelligence platform with 12 custom tools covering real-time market data, sentiment analysis, technical indicators (RSI, MACD, Bollinger Bands), portfolio P&L, and macro signals.

Real-Time Anomaly Detection & Crowd Monitoring System, UMBC Capstone | github.com/stutiupadhyay03/AnomalyDetection_Capstone | PyTorch, YOLOv8, ResNet18, CSRNet, OpenCV, Streamlit

• Production computer vision pipeline combining YOLOv8 object detection, anomaly classification (ResNet18), and crowd density estimation (CSRNet) — achieving MAE 31.18 on ShanghaiTech Part B benchmark. Deployed as live Streamlit app.

Healthcare Equity & Hospital Capacity in Maryland, UMBC Graduate Project (DATA608) | github.com/stutiupadhyay03/Maryland-Healthcare-Equity | Python, Pandas, Matplotlib, Seaborn, Statistical Modeling, Geospatial Analysis

• Designed and executed a quantitative epidemiological study across all 24 Maryland counties applying t-tests, ANOVA, multiple linear regression, chi-square, Fisher's Exact, and Kruskal-Wallis with full assumption testing (Shapiro-Wilk, Levene's, Breusch-Pagan) and effect size reporting. Identified statistically significant urban-rural disparity in licensed hospital bed capacity (p=0.008, Cohen's d=0.76); geospatial clustering mapped healthcare deserts across Western MD and Eastern Shore. Produced manuscript-style report with policy implications.

Spotify Streaming Analytics Pipeline, UMBC Graduate Project (DATA603) | github.com/stutiupadhyay03/Spotify-Streaming-Analytics | PySpark, Databricks, Spark SQL, GraphFrames, Pandas

• Analyzed Spotify's global Top 200 and Viral 50 charts (1.46 GB, ~9M records, Jan 2020–Dec 2021) using PySpark on Databricks — time-series streaming trends, regional performance, day-of-week behavioral patterns, and GraphFrames PageRank network analysis of artist collaborations.

Financial Anomaly Detection System, UMBC Academic Project(DATA690) | github.com/stutiupadhyay03/Financial-Anomaly-Detection | Python, Scikit-learn, Pandas, NumPy, Matplotlib

• Isolation Forest-based anomaly detection on historical stock data with Monte Carlo simulation (1,000 paths, 252-day horizon); top anomalies validated against COVID-19 crash and Federal Reserve emergency rate cut dates.

• Joint Secretary, Ananta 22 — Led strategic partnerships, designed participant communication frameworks, and orchestrated end-to-end event execution for 700+ participants. Built legacy management documentation adopted by future organizers.
• Overall Coordinator & Content Head, Spell of Cipher 2021 — Managed cross-functional teams, created newsletters and documentation for participants. PHP/MySQL competition portal with Rs. 20,000 in prizes.
• Overall Coordinator, Virtual Stock Market 2020 — Built a live interactive stock trading web simulation; designed participant policies and educational content.
• Design & Media Manager, Entrepreneurship Cell, GSFC University (2019–21) — Grew E-Cell from 5 to 11 members; organized entrepreneurship workshops and mentored students.
• AWS Community Day Volunteer (2022)
• Workshop Facilitator, 'Let's Talk Digital' — CODE Club, GSFC University (2020).

• Ranked 1st among 700+ teams — IIT Bombay National Entrepreneurship Challenge (NEC), 2019–20.
• 1st Prize, Group Discussion on Renewable Energy Solutions — GSFC University, Oct 2022.