Abstract
Artificial intelligence has rapidly transitioned from a futuristic novelty to a ubiquitous digital co-worker, automating tasks ranging from report generation to complex code development. While these advancements do promise unprecedented gains in workplace efficiency, they also simultaneously introduce significant technical and social challenges. On a technical level, the development of sophisticated software, including quantum computing simulators and decentralized communication platforms, increasingly relies on AI-assisted programming to manage growing complexity. Socially, the rapid integration of AI into professional workflows reshapes human agency, employee motivation, and the very meaning of work. The central challenge facing modern engineering is the pursuit of joint optimization: designing and implementing technical systems that maximize productivity without diminishing human experience or compromising the social fabric of the workplace. This portfolio addresses this dual-front challenge by exploring the social construction of productivity and job satisfaction in AI-driven environments while developing a decentralized communication system that prioritizes user privacy and resilient connectivity.
The technical project, "Mycelium: A Peer-to-Peer Chat System for Metadata-Minimizing Communication," involved the development of a decentralized, peer-to-peer (P2P) mesh network chat application. The primary objective was to create a resilient communication platform that allows users to discover peers and exchange encrypted messages without relying on centralized servers that could compromise privacy or act as single points of failure. Existing technologies, such as Telegram, Signal, Whatsapp, or Telemessage, all claim to provide superior encryption but have been either hacked, exposed to the government, or found selling user data to 3rd party companies. Our project stores zero metadata about the users, allowing for true decentralized connection. Utilizing a stack composed of Wails, Go, and the libp2p library, we implemented NAT traversal and hole-punching logic to facilitate connections across restrictive network environments. We also developed and hosted our own servers to facilitate peer discovery. Validation focused on performance and security, employing automated benchmarking and manual scripted walkthroughs. Results demonstrated a 100% connection success rate across various network scenarios, although discovery latency remained high, ranging from 23 to 106 seconds. Security protocols confirmed that message payloads are robustly encrypted both in transit via QUIC/TLS 1.3 and at rest using LevelDB ciphertext storage. Ultimately, the project successfully modeled complex P2P protocol theories in a functional GUI application, proving that stable decentralized communication is achievable despite relatively high initial synchronization times.
The STS research paper, "Balancing Efficiency and Humanity: How AI Adoption Shapes Workplace Productivity and Job Satisfaction," investigates the sociotechnical impacts of integrating AI into professional environments. The research addresses the "Productivity-Motivation Paradox," where AI-driven efficiency gains are often coupled with a decline in employee motivation and a sense of alienation. Drawing on Sociotechnical Systems (STS) Theory and the Job Demands-Resources (JD-R) Model, the study analyzes how AI acts as both a resource that enhances self-efficacy and a demand that can cause emotional fatigue. Through a qualitative analysis of recent scholarship and case studies, the paper argues that AI adoption is an underspecified process where the difference between empowerment and alienation lies in human agency. Findings suggest that when AI is used to augment human skills through AI delegation, satisfaction increases. Conversely, black box systems that reduce interpersonal interaction led to workplace loneliness and counterproductive behavior. The paper concludes that to achieve "superagency”, a state of empowered human-AI collaboration, organizations must move beyond raw efficiency to support social learning and preserve the expert identity of workers.
In the future, our technical project can benefit from integration with TOR, more user testing, and setting up more server nodes to facilitate faster connection. Together, these projects illustrate that the future of engineering lies not just in the robustness of the code, but in the intentional design of the social systems that the code inhabits. Whether optimizing peer discovery in a decentralized network or implementing generative AI in a corporate office, the goal remains the same: to create technical artifacts that empower individuals and foster meaningful human connection.
