Current Role
Founding Machine Learning Engineer, Dunedain
Working with the founding team the development of advanced machine learning and AI capabilities for a defense-focused startup. Building scalable ML pipelines, deploying real-time AI solutions, and collaborating closely with our highly talented team to deliver mission-critical technology for helping the Warfighter.
Selected MVP(-ish) Projects
BookFinder AI
Built a real-time book identification pipeline using YOLOv8 for object detection, integrated webcam capture, and Google Books API retrieval to match physical book covers to online metadata with high precision.
ELECTRA: Electrification Leveling and Efficiency Cost-Targeting ROI Analyst
Designed and built ELECTRA, an end-to-end platform for small-industrial electrification, combining ML-driven upgrade recommendations, interactive ROI simulations, and real-time energy profiling to empower data-backed decarbonization decisions.
Sentinal RAG
Built a RAG-LLM pipeline for real-time military support, integrating Whisper, FAISS, and GPT-4 to transcribe and embed open-source video data, delivering mission-relevant context and verifiable citations to the warfighter.
Real Time Anomaly Detection with a pyTorch Autoencoder
Developed a real-time anomaly detection system using a customizable PyTorch autoencoder pipeline, optimized for streaming data with threshold-based alerts and deployed on NYC taxi datasets for practical use-case benchmarking.
Former Ventures + Start-Up Experience
Elfur, Inc: AI Resume Matching
Pre-Seed Founding LLM/NLP Engineer
Led the design of LLM-powered resume summarization pipelines at Elfur, integrating MongoDB, OpenAI, Hugging Face, and LLAMA APIs to automate data workflows and reduce model training time by 300%.
Kero Sports
Post-Revenue-Pre-Profit ML Engineer
Deployed and optimized real-time event prediction models with 99.5% calibration accuracy, leading FastAPI-based GCP deployment that cut latency by 20% and boosted pipeline efficiency by 66%.
Optimal Dynamics
Post-Revenue-Pre-Profit AI Engineer
Built automated ML workflows and co-developed a deep RL supply chain model that drove $10M+ in profit, reducing analytics delivery time by 85% and leading data efforts a 5-person client-facing sales engineering team.
Academic Projects + Papers
Deep Learning for Satellite Orientation
Developed a deep reinforcement learning agent to autonomously and time-optimally orient satellites in 3D space using an Actor-Critic model trained in a simulated control environment.
Lexical Access Prediction from Semantic Embedding
Built a semantic embedding model to predict whether speakers access complex words directly or compositionally, advancing NLP applications in lexical cognition and morphology.
Multi-Indicator Computational Trading Portfolio Management
Built and optimized a multi-indicator machine learning strategy learner to outperform traditional benchmarks in stock trading, using adaptive window tuning and ensemble signal voting for dynamic portfolio decisions.
A Discrete Optimization Approach to Airline Flight and Captain Scheduling
Developed a binary linear programming model to minimize flight crew costs by optimally assigning captains to flights under FAA constraints, enabling scalable, regulation-compliant airline scheduling.