The scale of modern data centers enables developers to deploy applications across thousands of servers. The variety of applications and the scale of operations impose onerous challenge of meeting application performance requirements while maintaining efficiency. Today, data center operators typically over-provision the network and run services at low utilization to rein in latency outliers, thus decreasing efficiency. This large scale inefficiency results in high monetary, energy, and management expenses. This dissertation focuses on redesigning the end-host network stack to improve network efficiency and achieve low latency and latency variation at high utilizations . We begin by studying traffic emanating from modern servers across a variety of data center applications. We find that traffic is highly bursty, which contradicts the network flow model where traffic is uncorrelated. We use this observation to design networks that can benefit from bursty behavior. Second, in data center applications, predictability in service time and controlled latency, especially tail latency, are essential for building performant applications. Current practice has been to run such services at low utilization to rein in latency outliers, which decreases efficiency. To combat this, we present Chronos, which is a framework to reduce end-host latency and latency variation. Chronos reduces Memcached latency by a factor of 20 compared to typical deployments. Third, a range of new data center switch designs incorporating wireless or optical circuits depend on fast reconfiguration of the underlying topology. These hybrid designs assume a perfect, closed-loop control plane which end-host network stacks cannot provide today. We present the design and implementation of a closed-loop control plane using only software changes at the end host operating system that enable these topologies to support unmodified applications running over TCP. Taken together, these contributions demonstrate we can meet performance requirements of data center applications while running data centers at high levels of efficiency

The Notch signaling pathway is an evolutionarily conserved cell signaling system known to be involved in vascular development and function. Recent evidence suggests that dysfunctional Notch signaling could play a critical role in the pathophysiology of neurodegenerative diseases. We reviewed current literature on the role of Notch signaling pathway, and specifically Notch receptor genes and proteins, in aging, cerebrovascular disease and Alzheimer's disease. We hypothesize that Notch signaling may represent a key point of overlap between age-related vascular and Alzheimer's pathophysiology contributing to their comorbidity and combined influence on cognitive decline and dementia. Numerous findings from studies of genetics, neuropathology and cell culture models all suggest a link between altered Notch signaling and Alzheimer's pathophysiology. Age-related changes in Notch signaling may also trigger neurovascular dysfunction, contributing to the development of neurodegenerative diseases; however, additional studies are warranted. Future research directly exploring the influence of aberrant Notch signaling in the development of Alzheimer's disease is needed to better understand this mechanism.

A hallmark of human vision is the ability to rapidly recognize objects in a complex naturalistic scene. However, the exact mechanisms behind the computational invariance of object recognition remain unknown. In this study, we investigate object constancy by estimating how the ventral visual stream processes shading, shadows, textures, and specularities. To accomplish this, we use object meshes from the Objaverse dataset to create distinct multiclass classification tasks. For every task, we render a dataset by excluding exactly one of the previously stated features at a time. Subsequently, we train a ResNet50 model on each dataset. The trained model is evaluated on Brain-Score; deviations in these metrics indicate the importance of a brain region in achieving invariance to a specific feature. A reduced score for a removed feature in a particular region implies its crucial role in processing that feature since the model classifies objects based on remaining scene intrinsics.

This dissertation comprises three chapters representing a body of work examining the barriers that prevent individuals in developing countries from fully participating in markets such as labor and input markets and understanding how to address them. The first chapter, coauthored with Smit Gade, focuses on one such barrier: women's limited physical mobility and its impact on job search behavior. In many developing countries, factors such as safety concerns and restrictive social norms severely limit women's ability to move freely. As a result, women often require a travel companion to venture outside their homes, which can restrict their ability to search for and access employment opportunities if companions are unavailable. Coordinating travel with job-seeking women could help, but they may not know each other. We address this constraint in a field experiment in urban India. We match job-seeking women within neighborhoods and randomly vary whether they can coordinate their travel to factory interviews by scheduling them on the same date or different dates. Matching and coordinating travel increases interview attendance by 85%. The effects are stronger for women who knew fewer women at baseline and reported feeling unsafe when traveling. The treatment also improves job search beyond the interview experiment: women are 78\% more likely to visit prospective employers and make twice as many trips. Further evidence suggests that the effects are driven by women coordinating their travel, as matching without coordination has no impact. The second chapter, coauthored with Shilpa Aggarwal, Dahyeon Jeong, Naresh Kumar, David Sungho Park, Jonathan Robinson, and Alan Spearot, examines another important barrier to market participation: the role of travel costs in accessing subsidized agricultural inputs. Many countries provide agricultural subsidies but require farmers to travel to retailers to redeem them, much like regular market purchases. We study how travel costs affect subsidy take-up and input usage, particularly among remote farmers, using evidence from Malawi's Farm Input Subsidy Program (FISP). We find that travel-cost-adjusted input prices are substantially higher in remote areas, driven almost entirely by transportation costs. Nevertheless, subsidy redemption is nearly universal and only modestly lower in remote areas, suggesting that travel costs alone are not enough to deter redemption. Using a policy change that centralized beneficiary selection, we show that FISP significantly mitigates the steep remoteness gradient observed among non-beneficiaries. These results highlight how well-designed subsidy programs can help narrow spatial inequities in access to productive inputs. The third chapter, coauthored with Moumita Das and Anirban Sanyal, continues the focus on women's mobility by examining whether access to safe public transit can reduce barriers to labor market participation at scale. We study the rollout of the Delhi Metro Transit System using establishment-level data from India's Economic Census (1990–2013) and exploit spatial and temporal variation from the staggered opening of metro stations between 2002 and 2011. Using a Poisson Pseudo Maximum Likelihood (PPML) estimator in a two-way fixed effects event study framework, we improve upon standard models by implementing the Extended TWFE (ETWFE) approach proposed by Woolridge, 2021 and Woolridge, 2023, incorporating cohort-by-year interactions to account for treatment heterogeneity. We find significant and sustained gains in women's employment: within three years of metro access, the number of paid female workers per establishment increases by 0.95–1.59 (a 679%–1134% rise over a pre-treatment mean of 0.14), total female workers increase by 0.59–0.87 (278%–414% over a mean of 0.21), and the share of paid female workers among all paid workers rises by 2.2–5.4 percentage points (a 74%–180% increase from a 3% baseline). These results highlight the transformative role that safe, large-scale public transit infrastructure can play in easing mobility constraints and reducing gender gaps in urban labor markets.