Welcome to my portfolio

Hi, I'm Shruthi!

Software Engineer @ Microsoft

View Experience Download Resume
Shruthi R - Software Engineer

About Me

I'm a software engineer who gets genuinely excited about building technology that matters. Right now, I'm at Xbox working across a variety of cool projects; one such example being an artificial intelligence assistant for gaming that over 70% of users tried within a day of launch (still pinching myself about that one). Before gaming, I spent time making online spaces safer through intelligent moderation systems. Prior to that, I helped build cognitive health tools that can detect early signs of Alzheimer's up to 3 years before traditional methods.

What drives me? The intersection of complex technical challenges and real human impact. No seriously, whether it's reducing a gamer's frustration, protecting someone from online harm, or giving a family more time with a loved one, that's the work that gets me out of bed. When I'm not coding, you'll probably find me lost in a book (yes, I'm that person who insists books are always better than movies).

Technical Skills

Languages

Python C# Java C++ TypeScript JavaScript React HTML CSS

ML & AI

TensorFlow PyTorch Scikit-Learn Keras NumPy Pandas NLTK spaCy

Cloud & DevOps

AWS Azure Docker Kubernetes Terraform

Databases

SQL PostgreSQL CosmosDB DynamoDB Redis

APIs & Tools

REST GraphQL WebSocket Git GitHub Copilot

Work Experience

Oct 2024 - Present

Software Engineer 2, AI

Xbox Personalization

  • Led development of Game Assist, an in-game companion embedded in the Xbox Game Bar and PC app; architected the real-time context pipeline that fuses gameplay screenshots and live Xbox activity to deliver personalized, low-latency assistance, reaching 70% user adoption within 24 hours of launch.
  • Built Voice Mode for hands-free, in-game support, integrating speech recognition with multimodal context so players get real-time answers without leaving gameplay, cutting troubleshooting time by ~30%.
  • Engineered Highlight Reels, a background gameplay-capture system for the ROG Xbox Ally X that uses computer vision and gameplay signals to auto-detect standout moments and assemble shareable reels at session end, shipped to Xbox Insiders as Game Highlights (Preview).
Aug 2022 - Sept 2024

Software Engineer 2

Xbox Trust and Safety

  • Architected and deployed Small Language Models (SLM) to enable proactive content moderation, increasing detection accuracy by 65% across Xbox platforms.
  • Enhanced reactive moderation workflows, streamlining internal tools to boost efficiency by 18%, reducing human moderation workload.
  • Implemented proactive moderation strategies, mitigating spam and fraud threats by nearly 40%, safeguarding Xbox users.
May 2021 - July 2022

Software Engineer

Neurotrack Technologies Inc.

  • Engineered non-invasive cognitive tests leveraging AWS Microservices, improving early Alzheimer's and dementia detection accuracy, and identifying onset up to 3 years earlier than traditional methods.
  • Pioneered a coaching service to enhance cognitive health, resulting in a 15% improvement in client cognitive scores, positively impacting 100+ clients within the first 3 months.
  • Optimized backend infrastructure, improving scalability and reducing server response time by 25%, ensuring seamless user experiences.
Mar 2021 - May 2021

Research Engineer (Mitacs Accelerate)

Ethically Aligned AI

  • Analyzed literature across different domains in order to produce proof-of-concept for validating ethics in AI.
  • Iterated proof-of-concept upon feedback from a diverse class of businesses to identify specific business needs for building fair machine learning systems.
  • Strategized the implementation of a common tool for validating ethics.
Dec 2020 - May 2021

Software Engineer

Famwork

  • Leveraged statistical and ML algorithmic techniques for integrating machine learning models into electronic health record systems and built a chatbot for the patient portal.
  • Spearheaded the automation of the telehealth system for the minimum viable product, that was used to forecast diseases the patient was susceptible to acquire, based on his/her medical history.
  • Strategized the improvement of customer satisfaction by 50%, by automating the EHR system.
Sep 2019 - May 2021

Head Teaching Assistant

University of Alberta

  • Courses: Introduction to Foundations of Computing, Computer Networks, Graduate teaching assistant training.
  • Promoted to head teaching assistant in Spring 2020, prepared graduate TA training material and supervised 10+ TA's.
Summer 2018

Software Engineer

Makesto Infotech Private Limited

  • Launched a JavaScript plugin, which was used internally for supporting 3D models in the GL Transmission Format.
  • Established a framework for achieving dimensionality reduction of 3D models by 30%, with minimal data loss, in Python.
  • Collaborated with a team of 10+ people to maximize retail sales at Target by 80%, incorporating 3D modeling.
Summer 2017

Software Engineer

L&T Technology Services

  • Utilized agile practices and developed the login, sign-up, and deposit pages for the company using HTML5, CSS3, and JavaScript.
  • Participated in code reviews and contributed to building a complex, scalable software system.
  • Perfected the existing system by improving security against SQL injection attacks by almost 70%.

Education

M.Sc Computing Science

University of Alberta, Edmonton

B.E Computer Science

College of Engineering, Guindy, Anna University

What I've Built

Game Assist

Game Assist

Built the ultimate gaming sidekick for Xbox, an AI companion that knows what you're playing and understands your Xbox activity to provide personalized help. I developed features including Voice Mode for hands-free assistance during gameplay, screenshot-based context awareness for real-time support, and proactive coaching. Drove adoption by over 70% of users within a day of launch.

LEARN MORE: Xbox Game Assist | Xbox Wire Announcement
Highlight Reels

Highlight Reels

Built this gameplay capture feature for the ROG Xbox Ally X, enabling players to automatically clip their standout gaming moments, like epic boss battles and victories, into short replay clips to share with friends. I engineered it on top of Game Assist to capture clips in the background during play and assemble a highlight reel at the end of each session, requiring no additional downloads and never interrupting gameplay. Shipped to Xbox Insiders as Game Highlights (Preview) in the Xbox PC app.

LEARN MORE: Xbox Wire Announcement
Expense Tracker

Expense Tracker

Built a full-stack personal finance app that turns messy bank and credit-card statements into clear spending insights. I wrote custom PDF parsers that handle multiple bank formats (Amex, Scotiabank, Wealthsimple, Rogers Bank), an auto-categorization engine driven by layered user and global keyword rules, and interactive dashboards for monthly and category-level spending trends. Built with Python and Streamlit, backed by Supabase (Postgres) with user authentication and per-user data isolation.

USE: Upload bank and credit-card statement PDFs to automatically parse, categorize, and visualize personal spending trends over time.
Stock Market Predictor

Stock Market Predictor

A machine learning/deep learning project with a web interface used to predict the open price of a particular stock on a given day. Dataset created by scraping Yahoo finance. Created using Python.

USE: Used to predict the open price for a stock on a given day. Useful to set the highest price (Limit Buy) to buy a stock.
Hybrid UI

Hybrid UI - Towards Accessible Computing

A chrome extension for accessible computing. Improved user accessibility by 65%. Created using Python and tested on a control population of 30 people.

USE: Used to improve browser accessibility for people with low vision.
Knowledge Graphs

Relation Extraction in Knowledge Bases

A Natural Language Processing based project that focuses on relation extraction between a pair of entities in Knowledge Bases using a clustering algorithm. Created using Python and tested on KnowledgeNet relations.

USE: Can be used to identify and predict relations between a pair of entities, that can improve search results in Knowledge Bases.
Visual Aid System

Visual Aid System - Patented

A Visual Aid storytelling system used in story telling sessions for visually challenged children. Consisted of an interactive computer and sensor system and when scenes from stories were fed to the computer, the sensors rose to depict the scene.

USE: Enables visually challenged children to feel stories unfolding through their hands.
NEAL

Noise Eliminator for Active Learning - NEAL

An Interactive Machine Learning based algorithm that helps produce automated systems that provide highly accurate predictions, with a low number of training instances. Improved over previous systems by 10%.

USE: Can be used to improve predictions of any automated system and needs only a small subset of labelled data to produce stellar results.
Depression Detection

Early Detection of Depression Using Linguistic Metadata

A Machine Learning based project that detects the early onset of depression using the RSDD dataset and processed using Natural Language Processing. Achieved an accuracy of 99%.

USE: Can be used to detect depression in individuals and the diagnosed participants can be urged to seek professional help.
IoT Frailty Assessment

Platform to Assess Geriatric Frailty Using Smart Wearable Devices

An Internet of Things based system that assesses geriatric frailty using four smart devices which can be monitored from afar. Based on a smart condo environment and developed for Withings.

USE: Can be used to detect and assess frailty in individuals in real time.
Lung Cancer Detection

Early Detection and Prediction of Lung Cancer

A Machine Learning based project which focused on detecting and predicting lung cancer by automatically analyzing CT scans. Created using Python.

USE: Can be used to detect lung cancer, given a set of CT scans of the lungs.
WordPress Plugin

3D Models WordPress Plugin

A WordPress plugin to aid the upload of 3D models in the GLTF format, using the URL and file name.

USE: Can be used to render 3D models immediately on the web, with Lilliputian loading times.
Memotion Analysis

Memotion Analysis - SemEval 2020

Memotion (Meme + Emotion) Analysis is the task of analyzing sentiments from memes, typically posted in various social platforms, and classifying them into positive and negative sentiments. This system represents an approach to extract and analyze embedded texts of a meme from a released data set of a SemEval-2020 task.

USE: Can be used for sentiment analysis of memes across social media platforms.

Want to know more?

Like a one-pager? Here's my resume. Let's connect!

Download Resume