AI/ML Engineering Leader · New York Metro

Darian Garcia

I build and lead teams that take machine learning from research to production — from LLM research at Meta and Instagram ranking for 1.3B+ daily users, to computer vision running on satellites in orbit.

Get in touch LinkedIn ↗ GitHub ↗

Previously MetaInstagramHarvardPrincetonCapital OneCACI

10+: years shipping ML
1.3B+: daily users served
<100ms: on-orbit inference
4,500+: mission interventions

01 Experience

2025 — 2026
Senior Research Scientist, AI/ML

Meta · New York, NY
- Contributed core LLM research to MuseSpark 1.0–1.3 — model architecture, training efficiency, fine-tuning, and large-scale alignment.
- Led ML design of RecSys infrastructure powering Instagram Stories ranking for 1.3B+ daily users.
- Set multi-quarter technical direction across research and product teams in a large, matrixed organization.
2023 — 2025
Lead Software Engineer, AI/ML

Scientific Systems · Remote
- Architected a spaceborne compute platform — k3s and Docker on orbital satellites, bringing native Python and ML workloads to space.
- Led edge computer-vision models with a synthetic data pipeline for autonomous, real-time spacecraft identification and tracking.
- Quantized CNN/YOLO to embedded ARM via TensorRT/ONNX — 90% less labeled data at sub-100ms latency.
2022 — 2023
Senior Software Engineer, AI/ML

Capital One · Remote
- Invented first-of-its-kind load-balancer routing for card transactions, bypassing legacy mainframe rails.
- Deployed real-time credit-risk ML (XGBoost, streaming inference) as scoring microservices at sub-50ms latency.
- Built MLOps observability with statistical anomaly detection across high-volume transactions.
2019 — 2022
Senior Software Engineer, AI/ML

CACI · Remote
- Built the government's first end-to-end production ML pipeline for intelligence analysis, as customer-facing technical lead.
- Fine-tuned and deployed transformer NLP models (BiBERT, Multilingual BERT) for entity extraction at scale, with CUDA-level optimization.
2020 — 2021

Teaching Fellow

Harvard University · Harvard Extension School

Mentored the next generation of engineers across Python, modern full-stack architectures, and production-grade machine learning.
2016 — 2019

Earlier

UMBC — Research Scientist, AI/ML ThreatQuotient — Software Engineer Que Technology Group — Software Engineer

02 Expertise

LLMs & GenAI

LLM research and alignment, fine-tuning, language modeling, multimodal systems, synthetic data generation.

ML Infrastructure

Distributed training, inference optimization (TensorRT, ONNX), evaluation, MLOps observability.

Systems & Scale

Distributed systems, HPC, CUDA, Kubernetes / k3s / Docker, AWS, petabyte-scale data pipelines.

Leadership

Building teams, multi-quarter roadmaps, design reviews, mentoring, and customer-facing delivery.

03 Projects

Open source · Java

Minimal TAK Server

A security-first, lightweight server for the Team Awareness Kit (TAK) ecosystem with a deliberately small attack surface: protobuf-only Cursor-on-Target messaging, mutual TLS on every connection, Netty for TCP and Spring Boot for a REST API with health and metrics endpoints. Ships ready to run in a container.

View on GitHub ↗

04 Education

M.S., Artificial Intelligence4.0 GPA
UMBC
MBA, Information TechnologyExecutive leadership, strategy, and corporate finance
Quantic
B.S., Computer ScienceMagna cum laude in 3 years · 4.0 in major · Public-interest technology fellow
UMBC
Graduate Coursework, Software Engineering4.0 GPA
Harvard
Fine Arts & Humanities SymposiumSelected participant: ethics, policy, and the humanities
Princeton

AWS Certified ML Specialty · AWS Solutions Architect Associate · Microsoft MLOps Engineer Associate · Certified Scrum Master

Let's build something.

Open to conversations about AI/ML leadership, advising, and interesting problems.

hello@dariangarcia.dev