A product engineer specializing in production AI and AI infrastructure — from agents and pipelines to deployment, interfaces, and systems that scale.
A quick read on the scale and trajectory behind the work.
The principles that shape how I turn ambiguous problems into software people can actually use.
I enjoy difficult problems where the path from an idea to a useful product isn't obvious. My job is to understand the real requirement, make the trade-offs explicit, and then build something simple enough to ship and robust enough to keep improving.
AI and modern coding tools have made software development more accessible than ever. But getting something to production is a different problem.
Building a prototype is increasingly easy. Building something that remains reliable as it scales takes engineering judgment and experience — the ability to anticipate the inconsistencies, edge cases, security risks, compliance requirements, and seemingly insignificant details that users may never consciously notice, but that determine whether a product feels trustworthy or frustrating.
It means thinking beyond “does it work?” to “will it still work when people depend on it?” Can the system handle traction without becoming a liability? Can its data remain secure? Can the architecture withstand an audit? Can the product evolve without every change becoming a new source of failure?
That’s the gap I care about: turning what can be built into what can actually be trusted.
Understand requirements from the client's or user's perspective before choosing a technical solution.
Define acceptance criteria and a technical spec, including the scale the system may eventually need.
Add complexity only when the problem genuinely requires it.
Don't compromise on quality — but don't mistake 100% polish for the requirement to reach production.
Professional systems are shown at a high level where client work is private. Independent products show the work directly.
A code review tool that checks changes against an understanding of your entire codebase and prioritizes the suggestions that matter, not noise that gets ignored. Suggestion acceptance rate: 50–70%+.
Built and deployed an AI search experience that unified knowledge scattered across multiple enterprise applications into a single conversational interface, with citation-backed answers and deployment within the client's Google Cloud environment.
AI coding agents quietly eat your disk. Portée is a native macOS app that shows what Claude Code, Codex, Cursor and OpenCode are really costing your Mac, and lets you manage their MCP servers, skills, permissions and past sessions from one place.
A native macOS app for evaluating LLMs and decision models, optimized for decision models today with generation-model improvements coming soon. Test models and prompts, and build the datasets you need to fine-tune.
Designed a system that turns vague handwritten meeting notes into structured legal summaries, using available enterprise knowledge to fill gaps, asking clarifying questions instead of guessing, and improving through feedback.
A first-person maze game with 100+ levels, mirror mazes and spooky levels. Unlike the third-person maze puzzles out there, you are inside it - on all major operating systems, with VR coming soon.
A multiplayer platform that blends nostalgia with modern design, letting players enjoy fun new twists on classic childhood games like TicTacToe and Four In a Row.
A Flutter-based habit tracking app featuring home screen widgets, a hand-drawn aesthetic, incremental habit goals, and customizable day reset times.
Built the foundation for a new interaction model where users can operate an existing web application through an embedded conversational interface while retaining the traditional manual workflows.
A cross-platform Java desktop app that automated the tedious parts of playing Tambola (Housie) online during the COVID lockdowns, so everyone could join the fun.
A deliberately no-bloat Chrome extension whose entire job is to remove YouTube Shorts — from search, feeds, navigation, and direct URLs.
A three-year progression from intern to lead, with ownership spanning AI, product, engineering, delivery, and people.
Technical writing as another window into how I reason, learn, and communicate.
Three strong recommendations here; keep the full LinkedIn collection one click away.
Ankit Kishore approved these changes “I watched him go from AI Intern to Lead AI Product Engineer, and he earned every step. What sets Yash apart is ownership. He doesn't stop at writing code; he takes responsibility for the result from start to finish.”
Prajwal Tapsi approved these changes “He's a brilliant AI engineer with a rare ability to bring clarity to complex problems and turn them into practical solutions. What stands out even more is that he doesn't just move towards the goal himself, he brings the team along with him.”
Anoop Doddamane approved these changes “He was tasked with leading a team of ambitious engineers to create a successful delivery. His leadership skills and also his technical abilities to cater to the business requirements were something to highlight.”