● Developed a custom-colored Petri net execution engine for multi-agent systems with features like parallelism, batching, and logging.● Optimized agents for better loop control and execution state awareness.● Implemented state-of-the-art research for prompt and graph optimization in multi-agent systems.● Developed and implemented cloud-based CI/CD pipelines for unit testing, code analysis, and a custom patching system that streamlined maintenance of client branches, allowing tailored client-specific code without risking conflicts with the base code.● Enhanced multi-agent systems for use with less powerful LLMs by creating clearer prompts, improving information separation, managing token limits, and created tactics and strategies to aid with presenting information to the LLM.● Led backend development and micro-services creation.

● Developed a Python script to scrape over 300 million web pages, utilizing AWS EC2, andsubsequently stored the data in AWS S3 cloud storage.● Designed and implemented web-based applications utilizing FastApi, Django, and DjangoREST Framework incorporating sophisticated large language models (GPT-3/4) throughLangChain (LangServe).● Designed and implemented internal Django based web-console allowing team members toaccess and view currently running tasks, evaluate models (including different versions of thesame model) and manage running servers.● Designed a classifier proficient in efficiently detecting over 50 specific topics (genres) and theirassociated probabilities using Python, PyTorch, scikit-learn. Utilized G5 EC2 instances.● Designed and built custom langchain agents with chain-of-thought capabilities, supported bya constitutional agent responsible for supervising their activities.

● Developed game AI, including implementing pathfinding algorithms, decision trees, andmultitasking agents.● Engineered modular systems to support user modding, including the stat system, objectdefinition system, and parcels loading system, resulting in enhanced user customization andengagement.● Created a complex system to expand upon unity's API using reflection, function pointers anddelegates, allowing for increased coding efficiency.● Designed and implemented volume based (3D) fog of war and shadow casting system.● Designed an AI task override and priority system allowing for more complex behaviors.● Utilized ECS and DOTS to create parallel processing systems capable of processing largeamounts of data efficiently including pathfinding, fog of war rendering, and other backendsystems.

Qualified to ACM ACPC 2016Contestant at ACM SCPC 2016Contestant at ACM SPUC 2016