I'm a Security-focused Software Development Engineer specializing in ML infrastructure and distributed systems. I've delivered $4M+ in value by architecting enterprise platforms that reduced operational complexity by 50-97% across authentication, model deployment, and observability. I'm a proven leader driving cross-functional initiatives across 40+ teams with zero security incidents.
I graduated from Brown University with my Master's in Computer Science (2020) and Bachelor's in Applied Mathematics and Computer Science (2019).
Download my latest resume (June 2025) | Previous version
Building an AI-powered intelligence platform transforming how legal professionals conduct research and case strategy. Developed proprietary pattern recognition technology reducing research time by 50x while improving prediction accuracy. Led product and engineering from concept to pilot, integrating large language models with structured case law databases to deliver actionable insights. Read more here: https://www.briefcasebrain.com/
- Architected OpenTelemetry-based observability platform driving $3.2M quarterly revenue improvement through enhanced conversion rates. Reduced customer feedback processing from 8 hours to 3 minutes (97% improvement)
- Transformed data processing infrastructure with Rust-based compaction service, cutting costs by 63% ($1.2M annual savings). Reduced S3 throttling by 90% and automated recovery workflows from 5 hours to 5 minutes
- Designed enterprise IAM platform processing 8M+ daily authentication requests at 99.999% availability. Reduced dataset provisioning from 5 days to 3 hours through OAuth workflows, enabling 40+ engineering teams
- Streamlined ML model release cycle with Git-based artifact tracking, reducing deployment time from 3 weeks to under 2 weeks
- Led security audits for 15 services processing PII with zero high-severity findings. Created reusable frameworks that accelerated review cycles by 45% and reduced security incidents by 28%
- Developed privacy-preserving Continual Learning Platform with federated learning, reducing Voice Recognition model onboarding by 30%
- Conference Speaker: "Scaling Real-time Stream Processing for ML Pipelines" at Apache Flink Forward 2023 [Video]
- Teach AI fundamentals to high school students in the AI Scholars program, guiding them to build socially impactful AI projects in healthcare, climate, and social good domains
- Lead 1:1 Program sessions for AI + X Individual Research, providing personalized mentorship for students pursuing advanced AI research projects
- Design and implement GenAI applications for educational purposes, making complex ML concepts more accessible to students
- Work with small groups (5:1 student-teacher ratio) alongside experienced AI instructors from leading universities
- Guide students through practical applications of computer vision, NLP, and ethical AI, with many students going on to top CS programs
- Develop curriculum enhancements and mentorship frameworks that have helped scale the program's reach and impact
- Built and deployed machine learning models to production
- Created machine learning regression models and Docker container deployment infrastructure to Amazon Web Services using Fargate, ECS, Lambda, and ECR
- Built an ETL pipeline to evaluate machine learning models sales predictions for restaurants on AWS (DynamoDB, SQS)
- Visualized and analyzed time series model prediction performance using python
- Created a management system for a financial data framework using React, PHP, SQL, and Python
- Updated code deploy strategy and created dedicated server groups for applications
- Explored novel 3D scanning techniques to digitize Wayfair's vast catalog to be used for augmented reality, virtual reality and 3D rendering
- Developed a mobile scanning app that uses a depth sensor to generate a 3D scan of an object in C# using Unity
- Built a python testing framework for testing masking algorithms and photogrammetry workflow
- Created a file-sorting algorithm in python to increase 3D model generation efficiency
- Worked with 4 students to develop a Chrome Extension to hide harassing content on a variety of common social media platforms.
- Built a robust toxicity classifier that was trained on millions of comments using a Long Short Term Memory Model.
- Built a scalable flask-based API Endpoint served on AWS.
http://aanshshah.com/stop-harassment-site/
- Implemented multi-user networking in C# and created social virtual reality application in C# using
Unity for the GearVR and Oculus Rift.
- Created an animated navigation experience for the Oculus Rift using IBM's Watson Speech APIs and
Geolocation APIs using Unity in C#.
- Prepared and delivered weekly presentations on new technology and trends in the virtual reality space.
- Founded NLP-based candidate screening platform processing 1000+ candidates monthly
- Built telephony SDK reducing partner integration from weeks to days, improving candidate quality by 40%
- Used Twilio API to record automated phone call interviews and created natural language processing algorithm to prioritize candidates
- Raised thousands of dollars in seed funding
- Developed a video-based networking platform used by more 500 users with Django and Javascript.
- I designed and built an instant messaging platform that was used by over 200 active users who generated more than 7,000 messages.
- Created software to generate thumbnails upon video upload and produced 10,000 thumbnails over the course of the summer.
- Mentored students to develop novel solutions for data science problems
- Created automated grading infrastructure for the class using python and helped develop class material for a class size of 140.
- Developed assignment on searching algorithms (breadth/depth first search, bidirectional search, iterative deepening search, A*)
- Helped make assignment on knowledge representation and reasoning in Prolog.
- Hold weekly office hours and answer students' questions about assignments and course material.
- Managed 18 teaching assistants and handled grade complaints as Head Teaching Assistant (Fall 2019)
- Managed 7 teaching assistants, developed course materials, held office hours, and graded assignments for a class size of 40.
- Designed course from scratch with Professors Daniel Potter and Sam Watson
- Created assignments on Convolutional Neural Networks, Recurrent Neural Networks,
Attention/Transformer Models, and other Deep Learning topics using Keras and Tensorflow.
- Used numpy and sympy to create models for differential equations and linear algebra algorithms.
- Worked with Professor Vladimir Dobrushkin to create an easily accessible scripting tutorial for
students.
Notable Awards: 1st place (APA) at Intel International Science and Engineering Fair, Finalist in New York State Championship Debate Tournament
We believe that video games are an essential part of a childhood. We aim to provide video games and consoles for the underprivileged youth in our community.
- Supervise and lead a team of 17 data journalists to produce infographics and write data-driven articles.
- Supervised the creation of news apps and interactive graphics.
- Led development of an innovative, interactive-platform to display data-driven articles: First Drafts
Associate (January 2017 - May 2017), Associate Director (May 2017 - December 2017), Director (January 2018 - December 2019)
- Collaborate on lesson plans with teachers and student coaches.
- Engage students in thoughtful debates about current policy issues.
- Responsible for generating tournament pairings and results.
+ Researched information on bills recently passed by the Assembly.
+ Drafted letters to constituents informing them about current legislation.
+ Attended local community events, press conferences and constituent meetings.
+ Participated in roundtable discussions with the Assemblyman, his Chief of Staff,
and Legislative and Communications Directors, to brainstorm solutions for current issues and ideas for
future legislation.
+ Helped design legislation: tax credits for alternative energy use and provide further safety for
state life guards.
+ Worked with environmental and health safety lobbyists.
+ Coordinated "Skintelligence" campaign that raised awareness about skin cancer.
- Used knowledge graph embeddings to capture common sense linguistic priors in reinforcement learning environments (with Professor George Konidaris)
- Co-author Publication: "Graph Embedding Priors for Multi-task Deep Reinforcement Learning" - NeurIPS KR2ML Workshop, 2020 [Paper]
- Working with Professor Michael Littman, Neil Sarkar, and Carsten Eickhoff on using unsupervised learning techniques to predict readmission within a 30-day window of a patient being released using a multi-layered neural network architecture
- First Author Publication: "Comparing Global with Disease-specific Machine-learned Readmission Prediction Models" - AMIA Clinical Informatics Conference, 2020. Improved ICU readmission predictions using XGBoost on MIMIC-III dataset [Paper]
- Working with Professor Michael Littman on socially-strategic autonomous driving using reinforcement learning and game theory.
- Employed an adaptive policy computer from the Stackelberg Equilibrium of a two-player driving game to enable autonomous cars to safely assert themselves in scenarios requiring social negotiation with a human driver.
- Publication: "Socially-Aware Autonomous Driving Using Stackelberg Games" [arXiv]
- Analyzed EEG and behavioral data from participants who have taken Tolcapone using targeted dimensionality reduction and regression techniques
- Determined whether electrophysiological and behavioral signatures of enhanced prefrontal dopamine are consistent with it serving to stabilize prefrontal firing patterns and reduce state transitions
- Designed user interface for IFTTT that was used to aggregate user preferences for Trigger Action
Programming devices in Smart Homes and gathered data for device automation in Smart Homes.
- Coauthored submission to CHI 2017 Conference: "Mitigating Ambiguity in Trigger-Action Programming
for Home Automation."
UPOD
Study (N=259) explored the interaction between the gender of the bully and the victim, as well as the
impact of type of bullying behavior (physical versus verbal) on the perceived aggression of a bully
and the punishment given to him or her.
https://abstracts.societyforscience.org/Home/PrintPdf/12570
AWS Services: EC2, S3, Lambda, DynamoDB, Kinesis, Apache Flink, Fargate, ECS, EKS, Spark, CloudFormation, CDK
Security & Observability: OAuth 2.0, RBAC, OpenTelemetry, DevSecOps practices
Machine Learning: Federated Learning, Model Deployment, ML Infrastructure
Primary: Python, TypeScript, Java, Rust
Additional: JavaScript, C#, C, SQL, HTML/CSS
Apart from being a developer, I enjoy a variety of things. I enjoy traveling the world, playing golf, and binge watching shows on Netflix.
A comprehensive showcase of my technical projects spanning machine learning, AI, web development, and algorithm design. These projects demonstrate my expertise in building innovative solutions across various domains. All projects are available on GitHub
AI-powered medical analysis and prediction system for healthcare applications. Developing machine learning models for medical diagnostics and healthcare insights.
Implementation of linguistic priors in reinforcement learning environments for improved agent performance. Part of research on incorporating common sense knowledge into RL agents.
TensorFlow implementation of DeepMind's DQN with double dueling networks for Atari game playing. Includes advanced techniques for stable training and improved performance.
Final project for Deep Learning and Applications course, implementing advanced neural network architectures and training techniques.
ML model that generates articles and headlines based on Brown Daily Herald Opinion pieces. Uses advanced NLP techniques for coherent text generation.
Real-time Twitter sentiment analysis and visualization using D3.js and PubNub. Creates dynamic visualizations of sentiment trends across Twitter streams.
My personal portfolio website showcasing my professional experience, projects, and achievements. Built with responsive design and modern web technologies.
A searchable meme archive built with Django for efficient meme discovery and organization. Features advanced search capabilities and tagging system.
Highly customizable fantasy league platform with player creation, ratings, and league management. Features real-time updates and comprehensive statistics.
Navigate and explore a 3D space while interacting with text in an immersive environment. Creative experiment in combining 3D graphics with text interaction.
OAuth 2.0 flows for authenticating against the Spotify Web API in Node.js. Provides secure authentication and API integration examples.
Data visualization project analyzing the relationship between Walmart and Starbucks locations across different regions.
Implementation of algorithms to solve the vehicle routing problem for logistics optimization. Includes various heuristics and exact algorithms.
Implementation of the Davis-Putnam-Logemann-Loveland algorithm for boolean satisfiability. Efficient solver for SAT problems with optimizations.
Educational tutorials for using Python (SymPy, Matplotlib) in Differential Equations. Created for teaching mathematical concepts through code.
Implementation of sending emails through HTML forms using Google Script without a backend server. Useful for static sites needing contact forms.
Chrome extension that simplifies difficult words on webpages for improved readability. Makes web content more accessible by replacing complex vocabulary.
Amazon Alexa skill for tracking personal expenses through voice commands. Features natural language processing for expense categorization.
Fork of Gradescope's autograder samples. Examples of autograders for running on Gradescope's educational platform.
Fork of TensorFlow, Google's open source machine learning framework. Contributing to one of the largest ML frameworks in the world.
Collection of sample code snippets and examples in Java for various programming concepts and patterns.
Delivered conference talk on "Scaling Real-time Stream Processing for ML Pipelines"