I am currently interested in making large language models for code-related tasks more interpretable and trustworthy. My research lies at the intersection of causal inference, symbolic reasoning, and AI.
I'm especially interested in evaluating how large language models behave in real-world software engineering workflows such as debugging, bug localization, and code summarization. I study how causal frameworks can help explain or even repair language model outputs.
* Equal contribution
The course covered simplex method, duality theory, dynamic programing, and integer programming. Explored the modeling power of linear programming and its applications in supply chain, finance, marketing etc. We utilized the modeling language AMPL for building LP models.
The course covered algorithms for network flow problems focusing on the maximum flow, shortest path, minimum cost flow, minimum spanning tree problems. We used Python and the Python NetworkX package for hands-on practice implementing the algorithms covered in class.
The course covered models and tools to optimize production and distribution of products and services. In addition to traditional mathematical modeling, we explored the potential of reinforcement learning for supply chain problems.
Calculus-based probability class introduced Discrete Random Variables(Moment Generating Functions), Continuous Random Variables, Multivariate Probability Distributions, Sampling Distributions and the Central Limit Theorem.
This course coverse the following topics: Unix; Text editing; LATEX; tikz / xfig; Beamer; R programming language; Maple; APPL; C programming language.
Gained understanding of the concepts of Big Data and tools, including, MapReduce and Spark techniques, Hadoop, Scala.
This course introduced Machine Learning techniques and topics including: Linear Regression; Classification; Accuracy & Evaluation; Logistic Regression; LDA & QDA; Cross-validation, Subset Selection; Ridge & LASSO; Clustering; PCA/FA; Association Rules; Tree-Based Methods; Random Forests; SVMs and NNs.
Network location problems arise in many diverse applications. Examples include locating facilities, sensors, components, vehicles, people, services, and actuators. The course will include topics from classical location theory (covering, center and median problems) as well as more recent topics in the literature.
Topics include discrete-event, continous modeling apporaches, and fitting real-world data with distributions to create realistic models to simluate processes. Course culminates in a semester project in SIMAN - a high-level simulation language with C++/C interface.
The class focused on applications of statistical principles to empirical model building. Topics included Simple & Multiple linear regression - model selection and validation, diagnostics and remedial measures, detection of high-leverage observations, and robust fitting techniques.
Description: This 3-semester course covered the following topics: Limits, differentiation, integration; Polar coordinates, parametric equations, and series; Vector functions and their derivatives, partial differentiation, multiple integration, and vector analysis.
Description: Introduction to combinatorial analysis and graph theory. Topics include combinations, permutations and other counting methods, binomial and multinomial theorems, equivalence relations, graph theory, generating functions, and difference equations.
Description: First order equations; physical and geometric applications; Solutions of linear equations with constant coefficients; methods of undetermined coefficients; Application to network & dynamic systems; Introduction to series-solutions.
Description: Use of statistics for decision making, statistical description, frequency distributions, significance testing, sampling, hypothesis testing, statistical test using ANOVA and other statistical techniques as applied to business problems.
Description: Systems of equations, matrices, determinants, linear transformations, vector spaces and eigenvectors.
Description: An introduction to single-variable real analysis, the course covered sequences & series, the topology of the real line, limits, continuity, differentiation, & the Riemann integral. Emphasis on proof writing.
Description: Plant location, layout, and efficient operation. Includes practical applications of quantitative techniques such as linear programming, waiting-line problems, inventory control, and network analysis.
Description: This two semester course covered the following topics: Continuous and discrete probability distributions, random variables, limit theorems, stochastic processes, sampling, estimation and hypothesis testing; Multivariate probability distributions, hypothesis testing, linear regression, and statistical test using ANOVA.
Description: This course covers basic abstract algebra including groups and rings, and corresponding subgroups. Groups and subgroups, permutations, co-sets, direct products, homomorphisms, and factor groups.
Description: Introduction to the finance function of organizations and the long run decisions faced by firms. Valuation principles and present value techniques were developed and applied to securities prices and firms investment decision.
Description: The goal of this course is to study and understand issues facing firms in the global marketplace. Advanced course topics include exchange rate management through derivative instruments, understanding global financial markets, and investment evaluation and selection for multinational firms.
Description: Intermediate level treatment of the theory of price. Topics include consumer demand, production theory, factor pricing, and market structures.
Description: Firm’s investment, financing and dividend decisions are studied. Theories of value are considered under certainty and uncertainty. Recent developments and applications as included as needed.
Description: Equity securities and related markets are described from the perspectives of equity investing and portfolio management. Topics include equity valuation methods, mean variance theory, efficient markets, portfolio management and return measurement.
Description: Securities and the markets where they trade are described and evaluated from the perspective of individual investors and financial intermediaries. Topics include interest rate theories, financial intermediation, risk assessment, and fixed income security valuation methods.
Description: Intermediate level treatment of theories of national income determination and growth, business cycles and employment, inflation and the general price level.
Topic: “Monitoring and Progress in Human Rights: Investigating the Success of CEDAW.”
Description: Why do states vary in their protection of human rights? A great deal of cross-national quantitative research has cast doubt on the efficacy of international human rights law as a force for improvement in rights protections. However, one exception stands out to this trend: the Convention on the Elimination of All Forms of Discrimination Against Women (CEDAW). Several recent studies found that participation in CEDAW improves women's rights protections as measured by the Cingranelli-Richards Human Rights Data Project. In this study, we test the robustness of these findings for a CEDAW effect using a new dataset on the protection of women's legal rights created by the World Bank's Women, Business and the Law project. Looking at the period 1960-2010, we find that CEDAW membership is positively correlated with improvements for human rights, though this effect declines the longer a state is a treaty member. We also tested whether the CEDAW reporting process was correlated with periods of rights improvements in member states and found no evidence of this.Paper presented at the 112th Annual Meeting of the American Political Science Association, Philadelphia, PA. September 4, 2016.
Topic: “Hyperbolic Geometry.”
Description: The world we live in is considered Euclidean, which means that it is based on the Euclidean geometry. The Euclidean geometry itself is based on Euclid’s Elements, written circa 300 BC. The Elements is one of the most influential textbooks in history. Euclid derives all his results from the axioms stated at the beginning of The Elements, including the parallel axiom. The parallel axiom, also known as the fifth postulate, differentiates Euclidean and hyperbolic geometries because it does not hold true in the latter, also known as non-Euclidean geometry. Euclid’s ideas and Euclidean geometry, including the parallel postulate, have shaped our world. Thus, the concept of non-Euclidean geometry sets a foundation for exploring the world from an alternative perspective. In the hyperbolic geometry, we explored how various "truths" of geometry no longer hold and explore how this affects lines and triangles. The goal of this project was to learn about the development of hyperbolic geometry, what differentiated if from the Euclidean geometry, and how the hyperbolic geometric concepts had been used.
- Lead instruction and effectively maintain an exciting, engaging, and accessible classroom environment for a highly diverse group of 20 high school girls
- Manage and assess students’ progress in and proficiency of hard and soft computer science skills
- Manage 1 Teaching Assistant (TA) who serves as a support in classroom management, lesson delivery, logistical tasks, equipment management
- Lecturing for a 70-person calculus lab section and assisting students with conceptual understanding during weekly tutoring sessions
- Grade and provide feedback on weekly assignments
- Lecturing for a 35-person calculus lab section and assisting students with conceptual understanding during weekly tutoring sessions
- Grade and provide feedback on weekly assignments
-Graded and provided feedback on lab assignments and projects where students used Python programming language
Python: jupyter, matplotlib, numpy, pandas, scikit-learn, SciPy, seaborn, sklearn.
R: Amelia, CARET, class, dplyr, GGPlot2, naivebayes, rpart, vtreat, Zelig.
Basic Familiarity: Google Analytics, Hadoop, HTML, IBM Watson, LaTeX, MapReduce, Scala, Spark, Unix/Linux.
Scientific software: Maple, APPL, AMPL.
Probability, Statistics, Optimization, Math Modeling, Graph Algorithms, Data Wrangling, Data Visualization,
Machine Learning, Linear Algebra, Multivariate Calculus.
Russian, English, Turkmen.
My professional interests focus on building interpretable and trustworthy AI systems, especially for applications in software engineering. My recent coursework includes Deep Transfer Learning and AI for Software Engineering, further reinforcing my commitment to developing robust AI systems. Earlier graduate courses in Network Optimization, Data Mining, Probability Theory, and International Development Economics have also shaped my interdisciplinary approach. As an undergraduate student, I particularly enjoyed Differential Equations, Abstract Algebra, Money & Financial Institutions, and Equity Investment & Portfolio Management classes.
Book-wise, having graduated from a Russian high school, I have a deep appreciation for Russian classics: Pushkin, Tolstoy, Dostoevsky - you name it. In 2018, I discovered sci-fi by reading Isaac Asimov's Foundation series, which opened the door to new genres. More recently, I have been drawn to historical fiction, memoirs, and autobiographies that offer insight into the human experience across different cultures and time periods.
Despite the obvious lack of talent, I enjoy playing tennis and volleyball. I stay curious by learning and trying new things, even when success is uncertain. Recent personal wins include learning how to drive, becoming more comfortable riding bikes, and baking Nana's devil’s food cake. I also love traveling and discovering new places and cultures. Staying curious and grateful keeps me motivated every day.