Open to opportunities

ADITYA GARG

@adityagarg

Message

I build AI-powered web and computer-vision applications that ship real-time experiences.

India

Message

What I'm looking for

I’m looking for an entry-level role where I can build AI-native products—owning end-to-end features, working with real-time systems, and turning models into reliable, user-focused software.

I’m a BTech Computer Science student who builds AI-native software with a strong focus on architecture and real-time usability. I created an AI-Assisted Documentation Authoring Tool using Next.js, TypeScript, and ProseMirror/TipTap, where the document tree is the single source of truth and AI edits stream directly into tracked positions via Groq.

I also design low-latency systems and practical user workflows: I engineered a real-time AI Voice Receptionist using LiveKit and Deepgram with Groq streaming plus barge-in interrupt handling, and I built a Sports Tournament Management System with PostgreSQL bracket logic and WebSocket-driven live scoring. On the computer-vision side, I turned raw match video into rally-level analytics using TrackNetV2, OpenCV homography mapping, and YOLOv8, then packaged the results in an Electron/React dashboard with an integrated Llama-based Q&A. I enjoy owning end-to-end features, shipping dependable experiences, and iterating quickly based on how users actually interact with the product.

Experience

Work history, roles, and key accomplishments

Pepper & Pine

Built a real-time AI voice receptionist for restaurants with low-latency multilingual interactions using a dual-WebSocket bridge between LiveKit and Deepgram. Added Groq-based word-by-word streaming with interrupt handling (barge-in) and ElevenLabs TTS, plus a Next.js CMS backed by Supabase for menu, reservations, and AI configuration.

Python LiveKit Deepgram GROQ ElevenLabs WebSocket Bridging Barge In Handling Next.js Supabase CMS TTS

SOP Editor

Architected a custom ProseMirror schema and built a keyboard-navigable slash-command menu for SOP-style document editing. Implemented AI-native inline editing by streaming Groq LLM output into tracked document positions using structured JSON mapping to the schema.

Prosemirror Tiptap Next.js TypeScript GROQ Floating UI Slash Commands

ShuttleVision

Engineered a computer-vision pipeline to convert badminton match video into structured rally-level analytics using TrackNetV2 tracking, OpenCV court mapping, and rule-based shot classification. Built an Electron/React dashboard with heatmaps and trajectories, added a Firebase-backed match history system, and packaged the app as a distributable Windows installer with an integrated Llama 3.3 coachin

Python PyTorch YOLOv8 OpenCV Computer Vision TrackNetV2 Shot Classification Electron React Dashboarding Firebase Windows Installer

TopSeed

Architected a sports tournament management platform with PostgreSQL bracket logic and WebSocket-driven real-time scoring for concurrent live events. Implemented automated bracket scheduling, live leaderboards, and a RESTful API layer to unify workflows for organizers, teams, and spectators.

React Express PostgreSQL WebSockets RESTful API Design Leaderboards Bracket Logic API Integration

Education

Degrees, certifications, and relevant coursework

Manipal Institute of Technology

Bachelor of Technology, Computer Science

2023 - 2027

Activities and societies: Projects include SOP Editor (AI-assisted documentation), Pepper & Pine (AI voice receptionist), TopSeed (sports tournament management), and ShuttleVision (badminton match analytics). National-level badminton player; MIT Bengaluru badminton team captain (2023–2026).

Bachelor of Technology in Computer Science at Manipal Institute of Technology (July 2023–May 2027).