ML / AI Engineer • Computer Vision Engineer • Agentic AI Builder
M.S. Artificial Intelligence (Computer Vision) @ UCF • Graduate Research Assistant
Orlando, FL • Open to ML / AI / GenAI / Computer Vision roles
I design and build machine learning, AI, and computer vision systems that solve real-world problems.
My work focuses on:
- ML / AI Engineering: training, fine-tuning, evaluating, and deploying models with strong reliability
- Computer Vision: object detection, tracking, pose estimation, video analytics, and medical imaging
- Agentic AI / LLMs: RAG pipelines, multi-agent orchestration, tool-using agents, prompt design and evaluation
- Production Systems: Python services, FastAPI APIs, cloud deployment (AWS/GCP), CI/CD, Docker, monitoring
I enjoy taking ideas from research notebooks to production, with explainable behavior and measurable impact.
- Real-time computer vision pipelines
- Vision-language model reasoning and evaluation
- Agentic AI workflows and tool-using assistants
- Retrieval-Augmented Generation (RAG) systems
- Medical imaging and intelligent diagnosis platforms
- ML model training, tuning, benchmarking, and deployment
- Scalable backend services for AI applications
|
Problem: Automate complex desktop form-filling in a transparent, explainable way. Impact: Built a computer-use agent that turns UI perception into structured JSON plans and executes precise clicks and typing actions via a perception–reasoning–action loop.
|
Problem: Detect and track pedestrians across intersections for safety and planning. Impact: Deployed a YOLOv8 + DeepSORT pipeline over hundreds of scenes, achieving high tracking accuracy and enabling trajectory-based analytics for urban environments.
|
|
Problem: Measure and explain gender bias in multimodal reasoning models. Impact: Built an explainable pipeline using LoRA fine-tuned VLMs and Chain-of-Thought reasoning to classify bias with strong accuracy and human-readable rationales.
|
Problem: Turn structured family memories and lessons into personalized children’s storybooks. Impact: Designed a multi-agent GenAI platform where a root orchestrator coordinates Safety, Narrative, FactCheck, Pedagogy, Quiz, and Visual agents to produce safe, engaging stories with illustration prompts.
|
|
Problem: Classify multimodal WhatsApp messages into notify / digest / mute actions. Impact: Built an agent-style router combining metadata indexing, OCR, Whisper transcription, hybrid sparse+dense retrieval, rule-based safety filters, and confidence calibration to achieve high action accuracy on realistic message sets.
|
Problem: Build a unified medical imaging pipeline for multiple diseases with interpretability. Impact: Implemented a platform that supports diagnosis across five conditions, with Grad-CAM-style heatmaps to explain predictions to clinicians and support trust in model decisions.
|
Building intelligent systems that see, reason, and act.