Greater Chicago Area
• Designed and applied detailed rubrics for LLM evaluation, writing complex logic, math, and physics prompts to elicit reasoning failures and guide model improvement through process-supervision feedback. Performed red-teaming and fact-checking across STEM and general-knowledge domains, verifying claims with web sources and explaining weaknesses in model reasoning to refine future outputs. • Conducted large-scale preference labeling and comparative ranking of multiple LLMs (RLHF, DPO, RLAIF) while drafting critiques that articulated why certain completions failed rubric standards. Developed multimodal and tool-use evaluations integrating text, image, and retrieval tasks to measure grounding, reliability, and reasoning depth across diverse model architectures.
• Coach 10 clients to strive towards cognitive peak performance by understanding their unique learning styles and highlighting the importance of preventative health measures, resulting in an increased awareness and application of methods to elongate health span. Improve clients’ efficiency 5x by teaching methods to trigger and sustain flow, leading to improved performance in occupational tasks and more nuanced decision-making. Validate product ideas by understanding customer needs to prioritize features for an MVP. • Build continuous research systems to synthesize population-level trends to formulate preventative health solutions. • Conduct technical research to automate software processes while networking with software clients, enabling strategic alignment during discovery to identify pain points and create AI workflows to be included in a statement of work. Prototype end-to-end automation for product management. Enable LLMs to use internal application-level APIs resulting in 75% less human intervention for complex automated workflows spanning multiple platforms/applications. Implement a spec-driven, dynamic framework that extends Claude Code's capabilities by performing continuous deep research with skill development and curriculum discovery (inspired by Voyager, Eureka, Alpha Evolve, SPARC). • Design intricate digital twin simulations to prototype AI factories and industrialize entire feedback loops to implement systems-thinking. Implemented efficient, reproducible development environments (nixos, ubuntu, arch, k8s) for on-prem compute resulting in a reduction of operating costs by 50%. Automatically detect and offload simple tasks to local SLMs rather than routing to frontier models. Increase uptime after hitting token limits by rerouting messages in Claude Code through OpenRouter.
• Engineered a holistic performance optimization system to enhance cognitive and motor skills, applying an algorithmic feedback loop (Research → Implement → Analyze → Iterate) to systematically correlate and refine inputs, including sleep, diet, exercise, yoga, meditation practices, and specialized training protocols. • Reduce visuomotor reactions to 120ms from 220ms by using the ultralearning principles of directness and deliberate practice to optimize sensorimotor integration. Synthesize interdisciplinary research across cognitive neuroscience, peak performance psychology, and motor learning theory to design targeted subsystem protocols including stroboscopic vision training, visual flow state induction, and dynamic proprioceptive conditioning using a gyro ball. • Validate performance metrics within the top 0.1-1% of a 25M+ user environment (Kovaaks, Valorant, Apex) by executing a long-term data-driven analysis project to benchmark and optimize rapid skill acquisition; successfully reverse-engineered complex, dynamic systems to identify and eliminate bottlenecks.