Projects & UPCOMING Research
Guaranteed Architecture for Physical Security (GAPS)
Working at Dartmouth College’s Trust Lab with Sean Smith to apply Language Theoretic Security (LangSec) concepts to help strengthen network security of critical information systems.
[Long term] Explainable AI (XAI)
Interest in developing AI models with greater degrees of transparency, and human-interpretability that match the performance of methods such as deep learning.
Deeply exploring the design of a radical, new operating system through a project that is a culmination of my experience in theoretical computer science, HCI, design, and policy.
Notable features include: home screen that uses circle packing heuristics to efficiently display apps, an interface which naturally blends with the world around it using sensor data, along with an intelligent use of 3D graphics and AR throughout the OS.
Department of Defense
Applied my multidisciplinary experiences in collaboration with defense contractors and government customers towards developing next generational capabilities to support Department of Defense missions.
Computational Additive Combinatorics
Worked with Steven J. Miller at Williams College on a computational study the frequency distribution of random n-iterated sumsets, and explored the applicability of data-streaming framework towards studying massively large iterated sumsets.
Our results better characterize the behavior of the frequcency distriubution of in random iterated-sumsets as the probability of including each element, p, is varied. We also proposed a simple data-streaming framework for studying such random sets, and discuss the applicability of results in the field of data-streaming algorithms to future investigations into the behavior of random iterated sumsets.
Presented at Combinatorics and Additive Number Theory conference (CANT) 2019.
Hardness of Polynomial Local Search
Worked at the Technical University of Munich researching local search heuristics in a complexity theoretic context. Explored local search algorithms on H-minor free graph families, along with the applicability of relativization and algebraization to quantifying the hardness of PLS.
The main result involves a generalisation of a result of Scott Aaronson he coins ‘the transfer principle’ as it allows one to ‘lift’ results from communication complexity to algebraic query complexity. I use this result in combination with a very recent communication lower bound of computing the minimum of a boolean function given that its truth table is split across two parties to show that any proof of FP and PLS must not be algebraize.
Overall my results show that any proof that separates or collapses FP and PLS will require sufficiently advanced techniques (on par with those of P and NP). Which, in my opinion this lends some credibility to the opinion that FP is not equal to PLS (however this is not directly supported by these results).
Machine Learning Algorithm for Sensitive Data Identification
Worked with CHIMPS lab at Carnegie Mellon School of Computer Science to develop a machine learning based algorithm to identify sensitive data in natural language.
Given that NER has been studied in considerable detail, in both the context of formal and informal text, a core technique involved in the algorithm was to create a measurement of grammatical correctness to selectively well-tested NLP techniques.
MAJOR COURSE PROJECTS
ANALYSIS OF New-Type Warfare Attacks on America’s Agricultural SYSTEM
With the proliferation of automation through critical infrastructure subsectors the US is becoming increasingly vulnerable to asymmetric warfare tactics where our adversaries use non-violent means to slowly and subtly diminish our global influence. As a term project for Technology and Policy of Cyber War, I used my experience with the fundamental technologies driving this technological revolution to thoroughly analyze how precision agricultural technologies are making America’s agricultural sector vulnerable to asymmetric warfare tactics, and then further recommend concrete technical research directions which will be necessary to increase our resiliency to such tactics.
Participating in an interdisciplinary course-based program between the Tepper School of Business, School of Computer Science, and Heinz College of Information Systems and Public Policy to synthesize ideas from finance, technology and policy to create a proposal for a CMU-centric crypto currency.
We imagined using CMU Coin to further enhance how CMU students and faculty form work based connections. At the moment these connections are formed through media predating antiquity such as word of mouth, or printed posters. Our blockchain solution proposed using modern cryptographic primitives such as multi-party signature, secure multiparty computation (SMPC), and verifiable computing (VC) to facilitate advanced smart contracts which when paired with a campus focused incentive structure better facilitate student and faculty involvement in research projects.
Designed and studied image processing and machine learning algorithms to perform automatic quantitative analysis of biological data from labs.
Employed symbolic programming to explore and compare various probabilistic and automata based models for early-stage microscopic, vascular tumor growth.
Discrete Differential Geometry
Explored numerical and algorithmic challenges of translating ideas in smooth differential geometry into the discrete setting for computation on 3D geometry. Implemented algorithms solving problems related to discrete exterior calculus, curvature, conformal parameterization, and geodesic distance.
A BLOCKCHAIN BASED SOLUTION TO IMPEDE LABOR TRAFFICKING
Participated and won first place in Hacking4Humanity 2019 which focused on developing solutions to impede human trafficking or help human trafficking victims. We created an app called CleanWages which leverages blockchain contracts in a beautifully simple NLP-driven interface to help companies and job seekers engage in the hiring process safely and transparently.
MAKING financial literacy EASY AND FUN
Participated in a CMU challenge to combat financial illiteracy, and prototyped an app, Fintivity, that seamlessly integrates public educational resources with a motivational incentives structure inspired by Apple’s activity app.
Fintivity presents users with a list of short educational which introduce them to a general idea in finance. After watching the video, users gain access to more specific guides, along with a quiz to test the knowledge they learned.
Users can also set budgeting, spending, or educational goals, and view how close they are to achieving them in the Apple Activity app inspired rings. Finally users have the option to invite friends to compete with them.
Designed and programmed an intuitive and functional calculator with support for a range of operations far beyond the default calculator on the iPhone, specifically oriented towards discrete mathematics.
The app also supported multiple color schemes, and even night mode (at the time of writing the app, this was a unique feature).
[OLD] 3D Game Design
Created Black Horizon Games Urban Map Pack — a series of gritty and dark game environments based on famous ghost towns like Pripyat, the outskirts of Chernobyl, and abandoned winter villages. During development over five gaming websites wrote about my work.
Worked as a level designer on the popular Crysis mod, Casus Belli.
Designed concept images for State of Emergency, a Crysis 2 mod.
Placed 2nd in the Just Make Games Halloween Contest.