AI / ML Engineer
ML Engineer building production AI systems with measurable impact
Collaborated with NASA on applied ML research. Specializing in agentic AI pipelines, hybrid retrieval systems, and full-stack products that ship.
Flagship Projects
Scientists needed an interactive way to explore real-time oceanographic satellite data.
Led 5-person team to ship a live visualization platform, now exhibited at Kennedy Space Center.
Naive RAG systems hallucinate and give no way to measure or improve retrieval quality.
Hybrid retrieval pipeline raised Recall@10 from 58% to 81% with full query observability.
Exploring how LLMs reason across long-horizon decisions and simulate emergent behavior.
In development — multi-agent memory, planning, and behavioral divergence experiments.
About Me
AI/ML-focused computer science graduate with hands-on experience developing scalable systems and immersive interfaces for NASA and enterprise users. Expert in designing end-to-end AI pipelines, including React/TypeScript full-stack visualization layers, ImGui/OpenCV setup automation, and TensorFlow/PyTorch deep learning models, delivering quantifiable impact (e.g. 99% accuracy, 40% query speed-ups). Passionate about bridging technical innovation and user value to drive meaningful product outcomes in AI-driven organizations.
Resume Highlights
81%
Recall@10
RAGOps hybrid retrieval vs 58% dense-only baseline
NASA
PACE Satellite
Real-time oceanographic data visualization for NASA's PACE mission
5-person
Team Lead
Led cross-functional engineering team on NASA capstone
150+
Benchmark Queries
Human-labeled evaluation dataset for LLM system quality
Technical Skills
Experience & Projects
- Selected for a competitive 10-week internship with the Product Management AI/ML team.
- Contributing to the development and support of internal Agentic AI products to help automate customer service.
- Responsibilities include story writing, user testing, model evaluation, and cross-functional collaboration with design, engineering, and business units.
- Exposure to tools such as Azure DevOps, MySQL, Microsoft Copilot, Google Cloud Services(GCS), and LucidChart in support of digital transformation and intelligent automation efforts.
- Lead a team of 5 to develop a interactive data visualizations in React and TypeScript, enabling real-time analytics for NASA's PACE Satellite.
- Built full-stack applications integrating cloud data pipelines for large-scale scientific computations.
- Designed a Setup Wizard using ImGui and OpenCV, optimizing user configuration workflows.
- Implemented scalable APIs and GraphQL endpoints for cross-platform data integration.
- Currently being showcases at Kennedy Museum in Washington D.C.; Set to travel to different museums across the country.
- Built a production-grade RAG platform with hybrid retrieval (vector + BM25), cross-encoder reranking, and citation-based answer generation.
- Designed end-to-end LLM pipeline: document ingestion, semantic chunking, embedding, hybrid retrieval, and generation using FastAPI and PostgreSQL (pgvector).
- Implemented observability layer tracking retrieval latency, token usage, cost, and per-query diagnostics via an admin dashboard.
- Developed evaluation framework with 150+ benchmark queries measuring Recall@k, MRR, and answer correctness.
- Hybrid retrieval improved Recall@10 significantly over dense-only baseline; reranking reduced irrelevant context and improved answer precision.
AI Playground
Scenario
You have 1M labeled images and want to train a model to classify new photos into 1,000 categories.
"cat" attends strongly to "sat" — subject–verb relationship captured by the attention head.
Education
- In Progress
- GPA: 3.25 (4.0 Scale)
- Graduated
Get In Touch
Actively seeking ML Engineer and AI Product roles. Open to full-time and contract.
miked24977@gmail.com