# Portfolio Projects Source This source supports recruiter-chat answers about Atharva Fulay's independent and academic projects. It should stay aligned with the public Projects page when a project has a visible page section. ## Eff Tracking ### Problem and Goal Everyday work can disappear into a task list without showing whether effort is accumulating toward a meaningful weekly goal. Atharva built Eff to turn that work into a visible, reviewable weekly picture. ### What Atharva Built - A shipped iPhone and iPad product for setting a weekly Eff target, planning Today, and tracking completed effort. - Reusable Activities with weekday schedules and quick logging. - History charts for reviewing progress over time. - Partner and Group Circles for accountability. - Local-first sync, offline recovery, notifications, and account controls. ### Outcome and Availability Eff is shipped and active on the Apple App Store. It demonstrates end-to-end product ownership across planning, activity logging, progress review, accountability, offline behavior, and account controls. Public source material does not establish adoption or outcome metrics, so none should be inferred. ### Public Link - Apple App Store: https://apps.apple.com/us/app/eff-tracking/id6767980247 ### Useful Recruiter Questions - What shipped mobile products has Atharva built independently? - How does Eff turn recurring activity into visible weekly progress? - What product and reliability concerns did Atharva address in Eff? ## Automated Investing Agent System ### Problem and Goal Market research can require manually gathering news, evaluating signals, sizing positions, and monitoring trades. Atharva built the Automated Investing Agent System to test whether that workflow could be decomposed into specialized AI agents and automated with explicit governance. ### What Atharva Built - A multi-agent workflow separating research and portfolio-management responsibilities. - Integrations with external news and market-data sources. - Structured trade recommendations for review and execution. - Alpaca integration for paper trading rather than live-money deployment. - Scheduled execution and persistent storage. - Portfolio guardrails, including position-size limits. - An architecture that distinguishes research, reasoning, decision authority, execution, and governance. ### Outcome and Limits The system autonomously gathers information, evaluates opportunities, produces structured trade decisions, and executes them in a paper-trading environment. It demonstrates hands-on work with agent orchestration, APIs, persistent workflows, and safety controls. Investment performance should only be described with verified, date-bounded results. The source material does not establish a generalized investment track record, and the system should not be represented as financial advice or as live-money autonomous trading. ### Useful Recruiter Questions - What multi-agent systems has Atharva built outside work? - How does Atharva separate research, decision-making, execution, and governance? - What guardrails did Atharva use in an autonomous workflow? ## SMMRY-XT Browser Extension ### Problem and Goal Readers often encounter long, low-information articles and need a faster way to extract key points without leaving the browser. ### What Atharva Built Atharva taught himself HTML and JavaScript, integrated a summarization API, and shipped Chrome and Firefox extensions that summarized the active page, a URL, selected text, or pasted text. Features included: - configurable summary length and keyword count; - filtering sentences containing quotes, exclamations, or questions; - copying summaries and top keywords to the clipboard; - copying the current page URL; - extension packaging and distribution for Chrome and Firefox; and - a privacy model that did not track users, browsing history, or prior summaries. SMMRY-XT was an independent extension using SMMRY and was not the official SMMRY extension or endorsed by SMMRY. ### Outcome Atharva launched the extension in both browser ecosystems. It reached hundreds of weekly active users at its peak and demonstrates a long-standing interest in reducing information overload through practical software. ### Public Links - Chrome Web Store: chrome.google.com/webstore/detail/smmry-xt/oipkbgjabopdcfjbdjgablcjpjpoajci ## Duclade Collaborative Writing Platform ### Problem and Goal Collaborative storytelling happened informally in online threads, but existing products did not provide a structured way for writers to branch stories and contribute alternative chapters. ### What Atharva Built Atharva conducted user interviews, designed a content-subcontent-user data model, taught himself Groovy on Grails and Vue.js, built the backend and frontend, and deployed the product on AWS. The product supported a read, fork, write, and continue loop for branching narratives. After launch, Atharva continued interviewing users to understand engagement and retention barriers. ### Outcome Duclade attracted early interest but did not achieve sustained adoption. It remains a useful example of full-stack ownership, user discovery, cloud deployment, post-launch iteration, and an honest evaluation of product-market fit. ### Public Demo The archived Duclade demo remains embedded on the public Projects page. ## Nike Sneaker Analysis Atharva used StockX resale data to study sneaker-release hype cycles, resale volume, campaign timing, audience behavior, and marketplace dynamics. The project combined a release-timeline analysis, resale-volume exploration, revenue framing, notebook-backed visuals, and a personal Nike internship interview story. Public materials include the analysis notebook and release visual shown on the Projects page. ## Self-Driving RC Car Decision System For a graduate-school autonomous-driving competition, Atharva helped build an RC car that had to navigate an unknown course on limited Raspberry Pi compute. He designed modular software for image preprocessing, line detection, directional classification, and speed control. The team tested modules independently, used deterministic thresholds, and integrated them into a camera-based decision engine. The car completed the course, placed in the top 10, and finished within approximately two seconds of the top three while many teams did not finish. The project demonstrates early systems thinking, computer vision, modular design, and experimentation under resource constraints. ## Academic and Data Archive ### Mushroom Classification Decision trees and random forests used to classify more than 8,000 mushrooms with caution-weighted evaluation. ### Solar and Demographic Factors Project Sunroof, census, and IRS data used to explore relationships with solar adoption. ### Market Sector Analysis A Python workflow for scraping, API collection, modeling, and sector-level stock analysis. ### NBA Statistics Analysis Year-over-year player and team analysis using scraped Basketball Reference data and Tableau visuals. ## Project Themes and Tools Agentic workflows, multi-agent architecture, OpenAI APIs, Alpaca paper trading, scheduled execution, persistent storage, portfolio guardrails, HTML, JavaScript, browser extensions, REST APIs, Groovy on Grails, Vue.js, AWS, Python, computer vision, Raspberry Pi, modeling, web scraping, and data visualization. ## Useful Recruiter Questions - What agentic AI systems has Atharva built independently? - How has Atharva applied guardrails to automated decisions? - What products has Atharva shipped end to end? - What did Atharva learn from a product that did not sustain adoption? - What early projects demonstrate Atharva's systems and product instincts?