Soham Sunil Patil

Detail-oriented, organized, and meticulous employee. Works at fast pace to meet tight deadlines. Enthusiastic team player ready to contribute to company success.

1) AI‑Powered Insurance Underwriting
Technologies: Python, Flask, scikit‑learn (RandomForest), SHAP, Pandas, NumPy, Matplotlib, HTML/CSS/JavaScript

• Built a full end‑to‑end proof‑of‑concept web app for automated risk assessment and policy decisions, using Flask for the API and frontend demo.
• Engineered a synthetic dataset of 2,000+ insurance applications and trained a Random Forest classifier for “approve / approve with conditions / decline” decisions.
• Integrated SHAP explainability to surface per‑applicant feature‐importance (e.g., smoker status, age, BMI), including summary and waterfall plots (see figure).
• Designed frontend UI (HTML/CSS/JS) with interactive tabs for decision, confidence levels, key risk factors, SHAP explanations, and manual overrides.
• Persisted audit trail and override logs in CSV for downstream compliance and analysis.

2)Newspaper Content Prioritization Technologies: Python, Streamlit, Hugging Face Transformers (BART‑MNLI), NLTK (VADER), Tweepy, Plotly, Pandas

• Developed a Streamlit web app to help sub‑editors rank incoming articles by priority‑score, combining recency, sentiment (VADER), Twitter engagement (retweets & likes), and customizable category/location weights.
• Integrated zero‑shot classification (facebook/bart‑large‑mnli) to auto‑categorize articles into Business, Sports, Technology, Politics, or Entertainment.
• Fetched real‑time Twitter metrics via Tweepy, normalizing and weighting retweets/likes to compute engagement scores.
• Visualized analytics with interactive Plotly charts and data tables: priority‑score distribution, category breakdown, and filters/sorting.
• Enabled CSV export of full reports and priority‑score summaries for downstream analysis and archival.

• Certified AI/ML Intern, GennieSphere - September 1, 2025 to November 30, 2025.

• Worked in a team of three to develop the project “Newspaper Content Prioritization” as part of an apprenticeship program at Haptiq.