Jetlir Duraj (read as yet-leer doo-rai)
Jetlir Duraj (read as yet-leer doo-rai)
Contact: <first name><last name><at><gmail><dot><com>
Research Interests: Machine Learning, Search Engines, Quantitative Finance, Probability Theory
I have worked on Economic Theory/Mathematical Economics in the past.
Currently employed in San Jose, CA as a ML researcher.
Current research:
Proprietary and employment-based, mostly involving applications of ML to e-commerce.
Research:
Applied ML:
A Chain-of-Thought Approach to Semantic Query Categorization in e-Commerce Taxonomies, with Ishita Khan, Kilian Merkelbach and Mehran Elyasi, oral presentation in e-Comm workshop SIGIR 2025
Quantitative Finance and Machine Learning:
Black-Litterman End-to-end, with Chenyu Yu -- code, slides
presented at the Society of Quantitative Analysts, NYC
Deep Learning for Corporate Bonds, with Oliver Giesecke -- code, slides
presented (by coauthor) at the Advanced Financial Technologies Lab, Stanford University
Financial Math:
A multi-agent targeted trading equilibrium with transaction costs, with Jin Hyuk Choi and Kim Weston -- SIAM Journal on Financial Mathematics, Vol 15, No 1 (2024) pp: 161-193
Probability Theory:
Invariance Principles for Integrated Random Walks Conditioned to stay Positive, with Vitali Wachtel and Michael Bär -- Annals of Applied Probability, Vol. 33, No. 1 (2023) pp: 127-160
Martin Boundary of Random Walks in Convex Cones, with Kilian Raschel, Pierre Tarrago and Vitali Wachtel -- Annales Henri Lebesgue, Vol. 5 (2022) pp: 559-609
Invariance Principles for Random Walks in Cones, with Vitali Wachtel -- Stochastic Processes and Their Applications, Vol. 130, No. 7 (2020) pp: 3920-3942
On Harmonic Functions of Killed Random Walks in Convex Cones -- Electronic Communications in Probability, Vol. 19, No. 1 (2014) pp: 1-19
Random Walks in Cones: The Case of Non-zero Drift -- Stochastic Processes and Their Applications, Vol 124, No. 4 (2014) pp: 1503-1518
Green Function of a Random Walk in a Cone, with Vitali Wachtel (2018)
Mathematical Economics, with an occasional flavor of Operations Research:
Dynamic Information Design with Diminishing Sensitivity over News, with Kevin He, Theoretical Economics, Vol. 19 (2024), pp: 1057–1086
Identification and Welfare Evaluation in Sequential Sampling Models, with Yi-Hsuan Lin -- Theory and Decision, Vol. 92 (2022), pp: 407-431
Costly Information and Random Choice, with Yi-Hsuan Lin -- Economic Theory, Vol. 74 (2022), pp: 135-159
On Expected Utility in the Optimal Stopping of Diffusions -- Operations Research Letters, Vol. 48, No. 6 (2020), pp: 811-815
Bargaining with Endogenous Learning (2020)
Dynamic Random Subjective Expected Utility (2018)
Education:
B.Sc. degrees (Mathematics, Economics) and M.Sc. degrees (Mathematics, Economics) from Ludwig-Maximilians-University of Munich
Ph.D. in Mathematics from Ludwig-Maximilians-University of Munich, Germany
My thesis in Mathematics proves several structural theorems, called invariance principles in Probability Theory, for certain conditioned stochastic processes in discrete time called conditioned random walks. The oldest, simplest and most well-known invariance principle out there is the Central Limit Theorem! Within mathematics my results have found application, among other things, in enumerative combinatorics (see e.g. Courtiel et al, Melczer and Wilson, Borga and Maazoun, Borga), in the study of random planar maps (see e.g. Kenyon et al, Gwynne, Holden, Sun (1), and Gwynne, Holden, Sun (2)), and in graph theory (e.g. the recent Donderwinkel, Kolesnik)
Ph.D. in Economics from Harvard University, Cambridge, MA
My thesis in Economics focused on Economic Theory/Mathematical Economics. I produced three papers on Stochastic Choice (1, 2, and 6), a subfield of Axiomatic Decision Theory, two on Behavioral Economic Theory related to Mechanism/Information Design (4 and 7), two on Optimal Stopping (3 and 8), and one in Bargaining Theory, a subfield of Game Theory. My so far most cited paper out of these, is about finding axioms/conditions on choice data that characterize the preferences and private information of a dynamic agent in dynamic situations. Although it is not a paper related to the inverse reinforcement learning academic literature, if you have heard of IRL, this may ring a bell!
Of all the degrees, the Ph.D. in Mathematics is the one I am proudest of.