John Chester
New York, NY · john.anthony.chester@gmail.com
linkedin.com/in/johnchester212 · GitHub: @JohnAChester
Education
Fordham UniversityNew York, NY
Master of Science in Data ScienceGPA 4.0
Graduate Research Assistant for Economics of Cybersecurity2025 – December 2026
University of Maryland, College ParkCollege Park, MD
Bachelor of Science in EconomicsGPA 3.25
Specialization in Econometrics, Statistics, and Behavioral Economics2017
Technical Skills
- Programming Languages: Python, SQL
- Data Analytics & Tools: PyTorch, Scikit-Learn, HuggingFace, NumPy, Pandas, Excel
- Machine Learning: Deep Learning (DL), NLP, Classification, Regression, Clustering, PCA, Hidden Markov Models
- AI: Dynamic Bayesian Networks (DBN), Markov Decision Processes (MDP), Probabilistic Reasoning, Logical Agents
- Statistical Modeling: Bayesian Statistics, Econometrics, Multivariate Distributions, Sensitivity & Scenario Analysis
- Generative AI: LLMs, Claude Code & Cowork, OpenAI Codex, Agentic Automation, Structured Prompt Engineering
- Data Visualization: Seaborn, Matplotlib, Tableau
Experience
Tick MittNew York, NY
Data Science & AI Engineering InternSummer 2026
- Designed, built, and shipped TiCK Heatmap; a live, community-powered tick-encounter and Lyme disease-prevalence tracking platform as sole engineer, spanning a serverless AWS backend (CDK, Lambda, Aurora Serverless v2, DynamoDB) and a React/Vite frontend; the product became the company’s central pitch for a $1M HHS TOPxHHS Tech Sprint grant that advanced to final-stage review
- Built an empirical-Bayes hierarchical Bayesian engine (Poisson-Gamma sighting-density and Binomial-Beta disease-prevalence conjugate models) with multi-level geographic pooling, design-effect weighting for within-submission correlation, and a tick-selection bias correction; enforced privacy-by-design with physically separated PII/incident data stores and zip-only geocoding
Social Media Management Agent
- Designed and deployed a low-code social media comment-management agent (SocBot: n8n, Google Sheets, Claude API) that fetches, analyzes, and drafts replies to 1K+ monthly comments across YouTube, Facebook, and Instagram, saving 15 working hours/month at under $10/month in inference cost
- Created automated weekly reporting using Python and GitHub Actions (cron job with failsafes), pulling data from Shopify and Amazon APIs and posting formatted summaries directly to Slack for weekly standups; used Plotly for trend visualizations
Automated Financial Dashboard
- Built Claude Skills loaded with data dictionaries and modular ETL scripts so the CEO could run ad-hoc analytics requests conversationally, producing interactive dashboard artifacts; saved 1-2 hrs/week of manual data collection while increasing flexibility
Customer Intelligence Insights
- Connected Zigpoll (post-purchase survey) and OuterSignal (customer profiling) via MCP; extracted the ten most common consumer sentiments from Zigpoll history (pointing to product skepticism post-purchase), informing a marketing education push
- Used OuterSignal’s customer-network data conversationally with Claude to develop business strategy, including a proposed expansion into the hospitality rural B&B market with 3-5 specific outreach targets identified
Projects
FinBERT Sentiment vs. Price-Based ML for Post-Earnings Prediction | CISC 5352, ML in Finance; with Patricia Angeles
- Built an NLP pipeline that pulled and classified ~27,000 text chunks from 480 Edgar SEC 10-K/10-Q filings (24 S&P 500 companies, 2021-2025) with zero-shot FinBERT and confidence-weighted majority voting to produce 2,400 filing-level sentiment labels
- Engineered 16 price-based features and trained four classifiers under walk-forward validation to predict 5-day post-earnings abnormal returns; found only 44.9% sentiment-price agreement (below the 50% baseline) and near-zero correlation (r = -0.043), consistent with the semi-strong Efficient Market Hypothesis
Terrain Traversability Estimation with a Dynamic Bayesian Network | CISC 6525, Artificial Intelligence
- Designed a dynamic Bayesian network (hidden Boolean traversability state, Gaussian wheel-slip sensor model, near-identity transition model) and implemented a per-cell recursive Bayes filter in Python/NumPy/SciPy to estimate terrain traversability from a mobile robot’s wheel-slip readings alone
- Built the full ROS/Gazebo pipeline — online slip estimation, lawnmower coverage navigation with laser-based collision avoidance, and heatmap visualization — validated on a TurtleBot3, correctly identifying low-traversability cells from a single pass
Comparative Analysis of ML Classifiers for Type 2 Diabetes Prediction | CISC 5800, Machine Learning
- Built a reproducible preprocessing pipeline over the CDC BRFSS survey (441K records, 330 columns), narrowing to 21 literature-grounded features and yielding 253K clean records for modeling
- Trained and compared four model families (logistic regression, linear/kernel SVM, PyTorch MLP), tuning the precision/recall trade-off via pos_weight and F2-driven early stopping to raise minority-class recall from 76% to 86% at a screening-appropriate operating point, with all four converging at ROC-AUC ≈ 0.82
Healthcare Experience
CignaNew York, NY
Business Analytics Advisor2021 – 2025
- Guided data-driven negotiation strategies for a $2.3B healthcare portfolio across key healthcare systems, physician groups, and ancillary facilities throughout New England, boosting Cigna’s competitive market position by 1.7%
- Engineered Python dashboards with data storytelling and customizable visualizations for high-stakes contract negotiations, driving savings, improving price transparency, and enhancing cost prediction; adopted nationwide
- Developed and modeled over 35 deals with hospital systems delivering $21M in savings above market expectations
- Leveraged patient and sales data using SQL to create dynamic models which increased planning efficiency by 20%
- Achieved more favorable contract terms in 65% of negotiations completed, for a projected savings of $6.4M
- Evaluated 8 hospital systems for Fee-For-Value programs, shifting 14% of dollars to a risk-sharing arrangement
Provider Contracting Advisor2019 – 2021
- Negotiated high-value healthcare contracts with leading hospital institutions including Mount Sinai Health System and Hospital for Special Surgery, both strengthening Cigna’s network and restructuring existing agreements
- Spearheaded surgery center renegotiation efforts in the NYC region, achieving $1.8M savings over one year
- Constructed and executed competitive analysis strategies for hospitals, securing favorable terms in 80+% of deals
AetnaNew York, NY
Project Manager2018 – 2019
- Managed over 10 projects with a team of 15, ensuring timely and efficient completion of each project, while leveraging advanced analytics to optimize processes and increase overall project savings by 15%
- Led the termination process, coordinating across Aetna to ensure regulatory-compliant communications to 350K+ patients