# Yash Ganar (Yash Sudesh Ganar) > Portfolio of Yash Ganar: Generalist across product strategy, frontend development, AI/ML engineering, and visual design. Site: https://yashganar.com ## Identity I am Yash Sudesh Ganar: Generalist, Builder, Strategist. Generalist across product, code, and visuals; also AI/ML engineering and creative direction. Central positioning: I plan products, build them with code, and market them with visuals, all by myself. I'm not the best at any one thing, but I'm dangerous at all three together. Most people need a team to go from idea to launch. I don't. Tagline: I build from obsession, not obligation. ## What I Do I move work from fixation to something that ships and lands: decide what should exist, build it in code, and carry it with visuals. The portfolio is built around that full loop, not a single specialty. Evidence across the Experience roadmap and Selected Work: product UX (AlgoBulls), RL research (IITRAM → ACM paper), generative AI (Kalakriti), RAG systems (LocalRAG), data quality (VIP), BD/sales with design support, game design storytelling (Smart India Hackathon), and leadership (Rotaract). ## How I Think My work is driven by obsession, not impulse. Every project begins with a fixation and ends with something that outlasts the brief. The Oxygen Framework is the explicit model for that, presented on the site as how I think about everything. ## Oxygen Framework The Oxygen Framework is how I think about everything: the site's philosophy section under that exact framing. Four linked stages (source: site philosophy carousel / lib/philosophy.ts): - Obsession (origin): All motion begins not with mild preference, but with fixation. Without obsession, stimuli remain noise. - Desire (value): From obsession arises desire, the moment an object transitions from being observed to being needed. - Action (momentum): Desire alone is inert. Only when it reaches sufficient intensity does it compel movement. - Outcome (reflection): Every action produces an outcome. Not merely a result, but a reshaping of future obsession. Closing: "To understand action, one must not ask what a person wants, but what they cannot stop returning to." Across product, engineering, design, and execution, the same loop holds: fixation decides what is worth attending to; desire turns attention into a concrete need; action is what ships; outcome reshapes the next fixation. That is why cross-discipline work stays coherent instead of becoming three separate hobbies. ## Product → Code → Visuals The site's about line and nav pillars (Product · Builder · Visuals) are one chain, not three silos: - Product defines what should exist: the problem worth solving and the decision a user should make. - Code (Builder) makes it exist: research systems, apps, and interfaces that actually run. - Visuals make people understand and care: motion, branding, and storytelling that carry the work past the build. - Oxygen connects them: the same obsession → desire → action → outcome cycle runs through product choices, engineering execution, and creative direction. I run that chain myself. Product case studies live on https://yashganar.com/product (SlopTeacher: https://yashganar.com/product/slop-teacher); Builder and Visuals are live. ## Selected Work ### LocalRAG Personal knowledge assistant: ingest markdown, hybrid retrieve from Chroma, stream grounded Gemini answers with MongoDB session history. Aim: understand why retrieval still matters when models already offer large context windows, not "another chatbot." Status: completed · September 2026 - Docs: https://yashganar.com/projects/local-rag - Markdown: https://yashganar.com/projects/local-rag.md - GitHub: https://github.com/yashterdayyyy/LocalRAG - Stack highlights: Next.js, Chroma, Gemini, MongoDB, Vercel AI SDK ### SlopTeacher Ownership-quiz platform for public GitHub repositories. Paste a repo, take a text interview, walk away with a pass, pending, or slop report. Tagline: Prove you wrote it, not the model. Status: shipped MVP · v1.1.0 (live product, not under construction) - Website: https://slopteacher.com - App: https://app.slopteacher.com - Case study overview: https://yashganar.com/product/slop-teacher - Case study Markdown: https://yashganar.com/product/slop-teacher.md - Product index: https://yashganar.com/product · Markdown: https://yashganar.com/product.md Case study map (HTML pages are public and crawlable): - Overview: https://yashganar.com/product/slop-teacher - Markdown mirror: https://yashganar.com/product/slop-teacher.md - Executive summary (on overview): Developers often struggle to explain the reasoning behind code in repositories they've worked with, making interview preparation difficult. SlopTeacher explores a way to turn a developer's own GitHub repository into an interactive interview experience. This case study documents the problem discovery, solution design, product decisions, and initial implementation of that experience. #### Product Discovery - Problem Discovery: https://yashganar.com/product/slop-teacher/problem-discovery - Problem Statement: Producing a working codebase does not necessarily mean understanding how it works. SlopTeacher assesses understanding of a public GitHub repository and returns a repository-grounded diagnostic. - Why This Problem Matters: Code production and ownership differ, technical confidence needs evidence, interview prep lacks personal context, and knowledge gaps are hard to spot alone. - Assumptions: Hypotheses to validate about overestimated understanding, interview prep gaps, need for objective evidence, deeper comprehension, and actionable feedback. No formal validation yet. - Research: https://yashganar.com/product/slop-teacher/research - Research Plan: Planned discovery research: objectives, questions, participants, analysis framework, assumptions A1-A8, success criteria, and limitations. Not yet completed. - Competitive Analysis: Landscape across four adjacent categories, seven competitor profiles, a directional feature matrix, competitive findings, and positioning hypotheses. - Market Research: Secondary signals from Stack Overflow, Octoverse, and HackerRank; segments and beachhead, JTBD hypotheses, sizing caveats, limitations, and claims to avoid. - User Personas: Early-career and interview-prep developers first, then independents, unfamiliar-repo contributors, mentors, and interviewers. - Jobs to be Done: Candidate JTBD statements for core assessment, interview prep, self-assessment, and sharing, plus critical questions to validate in interviews. - Key Insights: Ownership gap over code production; compete on structured assessment; interview-prep beachhead; badge unproven; primary research still needed. #### Product Development - Product Strategy: https://yashganar.com/product/slop-teacher/product-strategy - Product Vision: Enable developers to evaluate and demonstrate their understanding of a codebase through repository-specific assessments, emphasizing understanding over producing code. - Product Principles: Assess understanding not authorship; ground quizzes in the repo; make results actionable; be transparent about uncertainty; keep sharing optional; protect trust with clear boundaries. - Success Metrics: Measurement areas: activation, assessment quality, and continued value. Quantitative targets and internal performance data are not disclosed. - Product Requirements: Public GitHub ingest with compact snapshot, quiz length and difficulty options, text interview with pass / pending / slop, diagnostic report, optional badge, and auth. - Constraints: Public repos only, auth required, daily usage limits for sessions and model calls, and badges record a session result rather than authorship or competence. - Prioritization: MVP prioritized the core assessment journey; features outside that path were deferred to keep the product focused and operable. - MVP Definition: Shipped MVP v1.1.0: marketing site, authenticated assessments on public GitHub repositories, diagnostic report, and optional shareable badge and result page. The product is live. - Solution: https://yashganar.com/product/slop-teacher/solution - Information Architecture: Site map across marketing pages, the signed-in app (home, active session, sidebar), and the public badge share surface. - User Flows: Landing through interview, then results with or without a badge depending on whether the full test was finished. - Design Decisions: Locked product choices: landing refresh, GitHub paste CTA, bento results, optional README badges, and shipping UI on the testing branch first. - Prototype: Instead of a Figma prototype, we used an Agentic AI approach and deployed into testing branches before production for faster development. - Final Product: The live SlopTeacher product is available at slopteacher.com. - Reflections: https://yashganar.com/product/slop-teacher/reflections - SlopTeacher is live, but shipping it feels more like the beginning than the finish line. #### Go-to-Market - Status note: SlopTeacher is currently being deployed in live markets. Once event data is gathered, Go-to-Market results will be published here. - Overview: https://yashganar.com/product/slop-teacher (Go-to-Market tab) ### Autonomous Intersection Management Developed autonomous intersection algorithms using PPO and A2C reinforcement learning. Evaluated on collision avoidance, speed, and travel time, published with ACM. Began as IITRAM research internship work. - Paper: https://dl.acm.org/doi/full/10.1145/3700838.3703668 - Builder: https://yashganar.com/builder#intersection-rl - Stack: Python, PyTorch ### Kalakriti Kalakriti.ai: image-generative AI model. Led a multidisciplinary team at Nexathon Hackathon / Byte Benders, Rizvi College (Mar 2024); placed #5 of 50 teams. Documented on the Experience roadmap (no dedicated project page yet). Skills: Python, Generative AI, Leadership. ## Experience Chronological timeline (newest first). Source of truth: site Experience roadmap / lib/experience.ts. Includes roles and deliberate career breaks. ### Data Quality Analyst: VIP (Vermont Information Processing) (May 2026 to Present) Analyze and validate datasets to catch inconsistencies, support cleaning and verification, and work with teams to keep data integrity high. - Skills: Data Integrity, SQL ### Professional development: Career break (Feb 2026 to Apr 2026) Upskilled and earned certifications focused on Python, SQL, and JavaScript. - Skills: Python, SQL, JavaScript ### Product Development Intern: AlgoBulls (Jul 2025 to Jan 2026) Redesigned product UI for faster investment decisions and contributed to a major revamp with new strategy filters and improved usability. - Skills: Product, UX, Cross-functional ### Business Development Intern: AlgoBulls (Jan 2025 to Jul 2025) Handled 1000+ inbound calls, improved sales conversion by ~20%, contributed to 2× ROI, and supported digital presence with design and editing. - Skills: Sales, Communication, Design ### SIH Finalist, Game Designer: Smart India Hackathon (Nov 2024 to Dec 2024) Designed maps and storylines for a constitution-teaching game aimed at youth, combining exploration, minigames, and memorable visuals. - Skills: Game Design, Storytelling ### Career pursuit: Career break (Aug 2024 to Nov 2024) Focused on MBA CAT 2024 preparation as a deliberate career-break goal. ### Research Intern: IITRAM (Jun 2024 to Jul 2024) Developed and analyzed autonomous intersection management systems, testing reinforcement learning agents across environments for traffic efficiency. - Skills: Machine Learning, Reinforcement Learning ### Team Leader: Nexathon Hackathon / Byte Benders, Rizvi College (Mar 2024) Led a multidisciplinary team building Kalakriti.ai, an image-generative AI model, and placed #5 out of 50 teams. - Skills: Python, Generative AI, Leadership ### Vice President: Rotaract (District 3141) (Jul 2023 to Jun 2024) Led as Vice President of the Rotaract Club of Universal College of Engineering under District 3141, owning team coordination and club initiatives. - Skills: Team Management, Leadership ### General Body Member: Rotaract Club of Universal College of Engineering (Jun 2022 to Jul 2023) Served under the Professional Development avenue and contributed across club activities, building early leadership and collaboration habits. - Skills: Teamwork, Professional Development ## Capabilities ### Product & strategy Product strategy, UX, interface redesign, cross-functional collaboration; PRDs and user research called out on specialised Product resume. ### Design tools Canva, Figma, CapCut, Premiere Pro, DaVinci Resolve ### Development tools Python, Cursor, Claude Code, GitHub (also JavaScript/TypeScript, Next.js / React across the site) ### Technical tools MongoDB, ChromaDB ### Deployment tools Vercel, Upstash Redis, Neon, PostHog, Sentry ### AI & ML RAG, Prompt Engineering (PyTorch used on research / Autonomous Intersection work); systems with Chroma, Gemini, MongoDB, Vercel AI SDK ### Design craft Branding, typography, UI/UX design, thumbnail, poster design (After Effects used on ROOP motion work) ### Soft skills Storytelling, research, public speaking, presentation, content strategy ## Visual / Creative Work Home skills and Experience also cover branding, typography, UI/UX, thumbnails, poster design, game design / storytelling (SIH), and design/editing support (AlgoBulls BD). ROOP (https://yashganar.com/visuals): "a visual restaurant." Every dish is a video, cut to be tasted in seconds, served course by course. Current dish (Amuse-Bouche): DaVinci Flashcut: sequence-cut flash animation for eye-catching hooks (DaVinci Resolve ~90%, After Effects ~10%; Tracery effect). Instagram: https://www.instagram.com/reel/DVUz33zgnRh/ Later courses (Appetizers → Digestif) are still empty on the menu. ## Writing Short pieces on building, fear, and judgment (https://yashganar.com/blog · https://yashganar.com/blog.md). - Me vs Me: I stopped writing my own code and called it a time problem. It was fear of what my hands would say. https://yashganar.com/blog/me-vs-me · https://yashganar.com/blog/me-vs-me.md - The world is getting more towards actions than theory: The gap that used to separate knowledge from execution is closing. What remains is judgment. https://yashganar.com/blog/action-than-theory · https://yashganar.com/blog/action-than-theory.md ## Portfolio Navigation ### Contact - Email: yashganar90@gmail.com - LinkedIn: https://www.linkedin.com/in/yash-ganar-379b1621b/ - GitHub: https://github.com/yashterdayyyy - Portfolio: https://yashganar.com - Resume PDF (complete): https://yashganar.com/assets/resume/yash-ganar-complete-resume.pdf - Resume PDF (code): https://yashganar.com/assets/resume/yash-ganar-code-resume.pdf ### Preferred pages (HTML) - https://yashganar.com/: Home (about, Oxygen Framework, skills, projects, experience, contact) - https://yashganar.com/product: Product case studies (SlopTeacher live) - https://yashganar.com/product/slop-teacher: SlopTeacher case study overview - https://yashganar.com/product/slop-teacher/problem-discovery: Problem Discovery - https://yashganar.com/product/slop-teacher/research: Research - https://yashganar.com/product/slop-teacher/product-strategy: Product Strategy - https://yashganar.com/product/slop-teacher/solution: Solution - https://yashganar.com/product/slop-teacher/reflections: Reflections - https://yashganar.com/builder: Software engineering and research - https://yashganar.com/projects/local-rag: LocalRAG technical docs - https://yashganar.com/visuals: ROOP visual work - https://yashganar.com/blog: Writing - https://yashganar.com/resume: Resume downloads ### AI-oriented mirrors - https://yashganar.com/llms-full.txt: Fuller plain-text digest - https://yashganar.com/index.md: Home (Markdown) - https://yashganar.com/product.md: Product index (Markdown) - https://yashganar.com/product/slop-teacher.md: SlopTeacher case study map (Markdown) - https://yashganar.com/builder.md: Builder (Markdown) - https://yashganar.com/resume.md: Resume (Markdown) - https://yashganar.com/projects/local-rag.md: LocalRAG docs (Markdown) - https://yashganar.com/blog.md: Blog index (Markdown) - https://yashganar.com/blog/me-vs-me.md: Full post: Me vs Me - https://yashganar.com/blog/action-than-theory.md: Full post: The world is getting more towards actions than theory