Software Engineer (AI)Xsolla
Sep 2026 — PresentLos Angeles, USA
- Working on the payments infrastructure behind Roblox, Epic Games, Rockstar Games, Valve and Twitch
AaryanVedBhalla
01About
Welcome! I'm Aaryan, a sophomore studying Computer Science at the University of Waterloo. Currently, I'm joining Xsolla as an AI Engineer this Fall, where I'll be working on the payments infrastructure behind Roblox, Epic Games, Rockstar Games, Valve, Twitch and others. Previously, I was an AI Full-Stack Engineer at Decimal Technologies, where I built LLM agent pipelines with LangGraph and LangChain to automate multi-step workflows, and RAG systems grounded in live knowledge bases using pgvector and PostgreSQL.
On campus I'm at present a part of the Chassis Design Team for UW Formula Electric and have served as a member of the Undergraduate Residence Advisory Committee.
Outside of code my interests lie in Chess, Motorsport, Gaming and 3D-Design.
I am currently exploring the fields of Visual Intelligence, Hardware Development, Machine Learning, and Video Editing.
02Experience
AI Full-Stack EngineerDecimal Technologies
May 2026 — Aug 2026Delhi, India
- Built LLM agent pipelines in LangGraph and LangChain, replacing manual multi-step workflows with production automations
- Shipped RAG systems end to end: ingestion, embeddings, pgvector indexing and retrieval grounded in live knowledge bases
- Turned structured and unstructured operational data into queryable pipelines in PostgreSQL
Undergraduate Residence AdvisorUniversity of Waterloo
Sep 2025 — Apr 2026Waterloo, ON
- First point of contact for a residential community: conflict resolution, crisis response and peer mentorship
- Ran community programming alongside housing administration and fellow advisors
Founder & Executive DirectorSHISTECH
Apr 2021 — Apr 2025Delhi, India
- Grew a high school technology symposium from 1,100+ to 6,000+ annual participants across 40+ schools in India and Oceania
- Led a 70-member organizing team across IT, social media, design and marketing
- Structured a zero-fee model funded by CAD $450,000+ in sponsorship over four years, from Zebronics, GitHub, Replit, Taskade and others
- Built and maintained the entire technical stack as sole engineer: Node.js systems, registration flows and participant databases
Web Development InternStellar Information Technology
Jun 2023 - Aug 2023Delhi, India
- Built and maintained web application features in a production software environment
ICT Infrastructure - Lead VolunteerThe Sanshil Foundation
June 2023 - July 2024Delhi, India
- Designed and built an E-Library and ICT Labs across two branches, putting educational technology in front of underserved children
- Ran stakeholder discovery before specifying any technology, so the tools served real outcomes rather than assumed ones
03Selected Work
A Chrome extension and voice assistant that steers students toward secondhand and lower-carbon purchases, telling them what is safe to buy used and checking every AI suggestion against live Amazon listings so only real products at real prices are shown. 2nd place at PivotHacks.
A production AI agent system built on LangGraph, structuring multi-step reasoning as task graphs with defined triggers and routing logic, backed by a full RAG pipeline grounding LLM outputs in structured evidence and deployed on AWS with Docker and CI/CD.
A computer-vision system that measures hand-luggage dimensions from a single camera and flags oversized bags against airline cabin limits, with three swappable dimension-estimation backends and a confidence-aware compliance model that refuses to fail a bag it isn't sure about.
A full-stack, AI-native application that turns a Spotify library into something you can converse with, asking questions in plain language and getting answers grounded in your own listening history.
An NLP research analytics system that processes arXiv abstracts at scale, using sentence embeddings and similarity search to index, retrieve and score papers for originality.
A productivity platform folding tasks, habits and calendar into one surface, driven by natural-language conversation with an assistant rather than forms.
A healthcare screening tool pairing symptom analysis with computer vision to identify skin conditions, optimised for edge deployment on Raspberry Pi for low-resource settings.
04Contact