Allen Thomas

Distributed Systems, Systems for Gen AI, Topics in LLM Agents, Deep Learning

Reinforcement Learning

• Developing AI-powered code refactoring platform to address enterprise post-quantum cryptography compliance requirements, targeting organizations facing 2030+ migration deadlines.
• Built ML engine detecting 25+ cryptographic vulnerability patterns with 90%+ accuracy on test datasets.
• Developed AI-powered code analysis platform using LLMs and Ghidra decompilation, reducing manual code review time by 80% for post-quantum cryptography compliance.
• Implemented multi-language parser enabling analysis across 10+ programming languages, expanding compatibility from Java-only to enterprise-wide coverage.

• Raised analytics uptime from 99.0% to 99.99% and cut $250,000/yr by leading analytics infrastructure migration to AWS
• Migrated analytics pipelines from HBase to Redshift, enabling 10× traffic growth for 200+ customers with zero downtime
• Eliminated stream processing bottlenecks affecting real-time analytics by migrating legacy Storm infrastructure to Flink, reducing event latency by 35% for 40K+ support agents.
• Preserved 350+ TB of historical data during migration, maintaining 6+ years of customer analytics access
• Established Airflow standards and documentation across 5+ teams, saving 15 developer hours weekly
• Reduced ad-hoc engineering data requests by 40% by implementing Metabase self-service analytics platform
• Mentored 10+ hires on coding practices and system architecture, reducing time to first release by 35% compared to previous year.

Presented at Chicago Booth GNVC Finals - React Native, TypeScript, FastAPI, LLMs

• Shipped a react prototype to Global New Venture Challenge at Chicago Booth School of Business judges in under 3 weeks.
• Delivered production-grade backend with async workflows, Supabase persistence, and resilient GPS/camera upload flows
• Engineered image-based appliance recognition powered by LLMs, reaching 85% identification accuracy through image processing and structured prompt templates.
• Designed type-safe APIs with Pydantic and implemented real-time sync powering a multi-property maintenance dashboard.

LLM Steering, Behavioral Analysis

• Implemented control vector techniques to steer LLM behavior across personalities to modify model outputs.
• Built pipeline to analyze model responses at varying control strengths and evaluation methods for behavioral analysis.

Golang, Distributed Systems, Fault Tolerance

• Built and deployed HyDFS, a custom fault-tolerant distributed file system supporting replication, consistent hashing, and automated recovery, providing the foundational storage layer for distributed data processing.
• Designed a high-throughput, distributed stream processing framework with exactly-once semantics on top of HyDFS, enabling reliable real-time data handling across multiple nodes.

Autonomous Research Validation - Docker, Google Gemini, OpenAI

• Developed autonomous system that reproduced 12 out of 15 academic papers with zero human intervention, reducing research validation time from weeks to hours through intelligent Docker environment reconstruction.

IEEE Publication - Reinforcement Learning, Multi-Agent Systems

• Researched and implemented multiple reinforcement learning algorithms for multi-agent systems, developing a novel hide-and-seek simulation environment inspired by OpenAI research on Multi-agent Autocurricula