Kyran Flynn

Welcome to my portfolio.

I've studied and implemented automated decision making in both tech and academia.

Here are my top projects.

Kyran Flynn headshot

Multi-Intervention Sequential Decision Modeling

Built an end-to-end experimental pipeline to study reinforcement learning in sequential decision settings where multiple interacting interventions must be selected in combination at each timestep.

Demand Estimation for Dockless Bike-Share

Built an end-to-end machine learning system to estimate demand for dockless bike-share from censored usage data, supporting city planners on bike-share policy decisions.

Disturbance-Based Local Search for VRPs

Built a top-performing, large-scale vehicle routing problem solver using a disturbance-based local search approach I designed.

Dynamic Speech-to-Text Parameters for Alexa

Implemented reinforcement learning agents to dynamically set key search parameters in Alexa's speech-to-text pipeline and reduced latency by 30%.

Under Amazon NDA

Education

Columbia University

MS in Operations Research (PhD Coursework)
Obtained October 2025, qualified for PhD

Brown University

BS in Applied Math & Computer Science
Graduated with honors in May 2023

Experience

Applied Scientist (Security) Jan 2026 — now
Margin Research

Short-term internship in cybersecurity at Margin Research.

Doctoral Applied Machine Learning Scientist Aug 2024 — Dec 2025
Columbia University, IEOR

Completed doctoral coursework and earned an accelerated MS in operations research in the IEOR department, qualifying for the PhD. Conducted research under Prof. Lily Xu focused on decision-making systems for healthcare applications.

Built an end-to-end system for modeling and evaluating multi-intervention healthcare decisions, enabling intervention-level value estimation using reinforcement learning and mentoring a student contributor in its development.

Served as Head Teaching Assistant for a graduate machine learning course (70+ students) under Prof. Christian Kroer, holding office hours and creating the official exams.

Software Development Engineer Jan 2024 — Jul 2024
Amazon, PXT

Delivered production-grade full-stack features and built a data validation framework to safely support a 90% load increase for an international hiring software release.

Applied Data Scientist Jan 2022 — Oct 2023
Brown University, Applied Math

Built and deployed an end-to-end machine learning demand modeling system for dockless bike-share planning under Prof. Alice Paul, translating usage data into actionable guidance for local transportation policy, and received recognition as an INFORMS Undergraduate Operations Research Prize finalist in 2023 for an honors thesis in the applied math department on this work.

Software Development Engineer May 2022 — Aug 2022
Amazon, Alexa

Implemented reinforcement learning-based tuning in Alexa's audio processing pipeline, achieving a 30% reduction in speech-to-text latency and earning a full-time return offer.

Head Teaching Assistant Mar 2021 — Aug 2021
Brown University, Computer Science

Restructured the syllabus and built all assignments and exams for the course Discrete Structures and Probability under Prof. Michael Littman, leading a team of teaching assistants to support 100+ students.

Software Engineer May 2021 — Jul 2021
Pointz

Translated user needs into a routing optimization roadmap and redesigned the backend routing algorithm for bike navigation startup Pointz, reducing their routing time by 50%.