AI/ML engineer and M.S. Computer Science candidate (AI/ML specialization) at the University of Central Florida. I work on computer vision, multi-modal contrastive learning, and deploying models to edge GPU hardware (Jetson Orin AGX, Jetson Nano, DGX Spark).
- Currently: AI Solutions Architect at Command Post Technologies and Graduate Research Assistant in Dr. Lou's UNBOUNDED AI Lab (LLM serving and multi-agent orchestration)
- Studying: M.S. Computer Science, AI/ML specialization, UCF (2026–2028). B.S. Computer Science with minors in Mathematics and Physics, UCF (2023–2026)
- Based in: Orlando, FL
- Portfolio: jyler.dev · Résumé: jyler.dev/resume · LinkedIn: david-jackson-89135727a
| Area | Details |
|---|---|
| ML & computer vision | Object detection (YOLO), CLIP-style multi-modal contrastive learning, model evaluation and benchmarking, RAG and Light-RAG, LoRA/QLoRA fine-tuning, LangChain and LangGraph |
| Inference and serving | vLLM, SGLang, PagedAttention, FlashAttention; GPU inference profiling |
| Edge and accelerated compute | NVIDIA Jetson Orin AGX, Jetson Nano, DGX Spark; offline and air-gapped deployment |
| Languages | Python, C/C++, Java, TypeScript/JavaScript, SQL |
| Frameworks | PyTorch, TensorFlow, Keras, React, React Native, Flutter, Node.js, Express.js, Flask, FastAPI |
| Infrastructure | Google Cloud, Azure, AWS Bedrock, Linux, Docker, Kubernetes, PostgreSQL, Neo4j |
Closed-Loop Perception-to-Action Agent (independent, 2026–present) End-to-end vision-to-action agent for a partially observable real-time strategy game. YOLO-based perception trained on custom-labeled datasets, a two-tier control architecture (fast reactive policy plus coarse-cadence retrieval-backed reasoning), and a heuristic-vs-learned policy benchmark with the perception stack held fixed.
VITALS — Search-and-Rescue Agentic Base Station (Lockheed Martin–sponsored capstone, 2025–2026) Owned the mapping and planning agent in a four-agent architecture (vision, retrieval, reasoning, mapping/planning) on a six-engineer team. Re-architected the PostGIS/OpenStreetMap terrain pipeline, replaced a greedy allocator with an A* global planner plus local avoidance layer, and targeted a Jetson AGX Orin inside a 15–60 W power envelope. Repository · Case study
Kestrel — Autonomous Videography Drone (Blue Origin–sponsored, 2024–2025) Founded and led the project within ACM at UCF. Defined the system architecture across sensing, perception, and control on a modular ROS 2 stack; designed a MobileNetV3 + DeepSORT tracker (23.64% mAP, 35.34% Rank-1, 48.99% Rank-5); simulated a custom airframe with a 40 min best-case flight time. Repository · Case study
NewsMap.Today (2024) MERN news aggregator built around an interactive Leaflet world map; took the project from front-end developer to full-stack developer and project manager. Repository · API docs
Echo Interview (hackathon, 2025) AI mock-interview app: Gemini generates role-specific questions from a résumé and job description and gives real-time feedback. Led the AI integration and full-stack build in Next.js. Repository · Devpost
More at jyler.dev/projects.
- Lead Foundation Exam review sessions (F.E.A.S.T.) for UCF CS students covering C, data structures, and algorithms
- Give an annual introductory AI talk to a 3rd-grade class
- Volunteer with the Light Orlando Hunger Project