Tavi
Engineering standups are supposed to unblock people. In practice, every small question becomes a follow up: “what does that endpoint return?”, “can someone draft the ticket?”, “let’s just bump that timeout.” A team walks out with 15 tiny action items on top of the real work, and each one costs a context switch, a Slack thread, or another meeting. We’re building Tavi to clear those small items during the standup itself, so people only leave with the big tasks they own.
PARM
A causal benchmark for when an AI agent should use personal memory mid workflow. An agent works through a seeded tool environment, a late tool result makes one memory actionable, and PARM scores whether the agent retrieves that memory before it acts and stays silent when the cue is removed. 157 frozen scenarios, 471 executable cases, and deterministic scoring with no LLM judge.
gantry
A coding agent skill for Claude Code, Codex, and other .agents hosts that keeps engineers in the design loop while AI writes the code. Code generation is gated behind collaborative pseudocode the engineer must approve, and every decision is kept in a .gantry doc. Forward, continue, and rebuild modes cover new code, revisited designs, and auditing existing code. MIT licensed.
Repple
Repple is a competitive fitness app for iOS that makes working out fun and consistent. Users earn points for workouts, create streaks, compete against friends in weekly matchups, and climb leaderboards. Features include ELO matchmaking, automated point aggregation, and AI generated workout plans that adapt to user performance.
google-tools-mcp
An open source MCP server that unifies all of Google Workspace behind a single server and OAuth flow. 169 tools covering Drive, Docs, Sheets, Slides, Gmail, Calendar, and Forms. Features auto re auth, PDF/Word reading from Drive, and read before edit guards. MIT licensed, no telemetry.
Veritas
Veritas is a clinical research platform that combines cryptographic proof of personhood (World ID) with ML powered response quality scoring to give researchers clean, trustworthy data. Won Best Proof of Human Application at Catapult 2026.
Caladrius
Caladrius is a privacy first AI triage assistant. It aims to cut through the inefficiencies of triage intake and inequities in prioritization, while maintaining a privacy first HIPAA compliant ethos. Awarded 2nd place in HackGT 12's track for social impact.
Verbatim
Verbatim is an intelligent platform that takes any video, summarizes it for quicker consumption, translates it into multiple languages, and then recreates the speaker’s lip movements to match the new audio—delivering a seamless, localized experience.
FORMulator
FORMulator allows users to perfect any type of body movement on their own, by leveraging computer vision technology to provide instant feedback. The app uses a combination of pose estimation and machine learning to analyze the user’s movements and provide real time feedback on their form, helping them to improve their performance.
Photonic Implementation of QKD
Quantum Key Distribution (QKD) is a method of secure communication that uses quantum mechanics to exchange cryptographic keys. This project focuses on the photonic implementation of QKD with near infrared lasers, utilizing the principles of quantum mechanics to ensure secure communication between two parties.
Self-Organizing Maps × K-Means
This project explores how SOM, known for its capability to uncover topological structures, fared against traditional centroid based approaches in capturing intricate data relationships.
Quantum Racer
Quantum Racer is a simple car racing game that teaches players about quantum decoherence and noise by challenging them to navigate a quantum car through a track while avoiding obstacles. The game is designed to be fun and educational, making it suitable for players of all ages.
For startups and founding teams
Karthik Thyagarajan is a founder-engineer studying Computer Science and Artificial Intelligence at Purdue. At Samsung Research America, he co-architected an always-on AI agent runtime for XR glasses. He is currently building Tavi, a voice teammate for engineering standups, and co-founded buildpurdue, a founder community at Purdue. His work spans agent infrastructure, on-device AI, product engineering, and evaluation.
For an early team building AI products, his strongest demonstrated fit is owning a broad technical problem from product decisions through implementation and evaluation. His current interests are live voice and video AI, AI-first wearables, and infrastructure that lets agents discover and use internet services.
Substantial ownership of agent systems
At Samsung Research America (May to August 2026), he co-architected an always-on agent orchestrator in Kotlin and owned half of a 25+ module runtime. His reported on-device router achieved 95% signal suppression and 90% intent-routing accuracy. His work also covered tool permissions, model routing, memory, and phone control, connecting model behavior to a usable device experience.
Builds teams and operating systems around a problem
As buildpurdue's co-founder and co-president, he helped turn an informal gathering into a founder community with a selective cohort, weekly co-working, funding, and founder-led problem-solving workshops. He built much of its web platform and integrated CRM, applications, events, and check-ins. His account describes data on around 500 students and a mailing list of about 1,000.
Makes technical claims testable
He built PARM, an open-source causal benchmark for when an agent should use personal memory: 157 frozen scenarios and 471 executable cases, with deterministic scoring and versioned replay artifacts. Its portfolio account explicitly distinguishes development results from held-out claims. He also built google-tools-mcp, a published MCP server exposing 169 Google Workspace tools through one OAuth flow.
Pursues difficult product problems
He is currently building Tavi, a voice teammate intended to answer codebase questions and handle small action items during engineering standups. The project tackles multi-person turn-taking, when to stay quiet, sandboxed coding, and synthetic meeting evaluations. This is ongoing work, rather than a claim of shipped adoption. His published interests explain the direction behind it.
Co-founder who ships and thinks about distribution
At Repple (November 2025 to June 2026), he co-founded a competitive fitness app released on the App Store and built the entire FastAPI, PostgreSQL, and Supabase backend, including synchronized workouts, ELO matchmaking, and weekly competition logic. His firsthand account also describes marketing before the UI existed, campus distribution, and why local community density mattered to the product.
This brief summarizes Karthik's public portfolio and firsthand accounts. Employment contributions and internal performance metrics are self-reported; linked code and artifacts provide additional evidence where available.