FelipeArias
Felipe Arias
@FelipeArias
ML @ Uber | Founder @ Marovi AI
Location: San Francisco, CA · Org: Uber · felipefelixarias.github.io/
MLE on Uber's ML Training team. I build GPU training infrastructure, GenAI systems, and support LLM fine-tuning — spanning infra, applied ML, and production deployment.
Founder of Marovi AI, a research comprehension and translation platform that makes knowledge accessible across languages. Former NSF Graduate Research Fellow at UIUC.
Resume
Felipe Felix Arias felipefelixarias.github.io | felipefelixarias@gmail.com | GitHub | Scholar | LinkedIn SUMMARY Machine Learning Engineer working across ML training infrastructure, GenAI systems, and evaluation pipelines. Experienced across the full ML stack: distributed GPU training, model fine-tuning, agentic applications, and structured evaluation. Research background in programmatic data labeling and weak supervision. NSF Graduate Research Fellow; published with Google Brain researchers at IEEE ICRA. SKILLS Languages: Python, Go, SQL, PyTorch ML: DDP, FSDP, DeepSpeed, Ray, LoRA/QLoRA, XGBoost Infra: Kubernetes, Docker, AWS, Spark GenAI: Agentic systems, RAG, evaluation pipelines, Claude Code, Codex EXPERIENCE Uber Technologies — ML Training Team San Francisco, CA Machine Learning Engineer Mar 2024 – Present - Led migration of Uber’s ML training scheduler to a multi-region compute backend, moving thousands of production workflows with zero downtime. - Co-designed distributed GPU orchestration on Kubernetes for PyTorch and DeepSpeed workloads. - Built a multi-signal safety system in which GenAI-assisted signal analysis proposes flags and classification neural networks on telemetry validate them, driving measurable reduction in unsafe driver behaviors. - Designed a generation/decision separation pattern for safe LLM-in-the-loop production systems. - Built and deployed a RAG-based incident assistant that was broadly adopted internally, as well as an employee information agent. - Served as technical lead for LLM fine-tuning with external providers, advising multiple teams on fine-tuning workflows and evaluation. Marovi AI San Francisco, CA Founder 2024 – Present - Building a research comprehension and translation platform with a provider-agnostic LLM API spanning OpenAI, Anthropic, Google, and DeepL. - Built agentic translation pipelines with glossary enforcement, structured outputs, and crowdsourced correction loops across 9+ languages. - Reached the top 10% at Y Combinator twice. - Built evaluation infrastructure including benchmark frameworks using FLORES and XL-SUM, LLM-as-judge systems for automated quality scoring, and A/B comparison across providers. - Full-stack work across AWS, Python, Pydantic schemas, caching, and batching. University of Illinois at Urbana-Champaign — Parasol Lab Urbana, IL NSF Graduate Research Fellow Advisor: Nancy M. Amato 2019 – 2023 - Designed self-supervised programmatic data labeling pipelines in which simulation generated collision data at scale and spatial regression models distilled learned structure into planning heuristics, achieving a 5x speedup with zero manual labels. - Published with Google Brain researchers at IEEE ICRA 2021. - Released open-source simulation and evaluation tools for multi-agent datasets. - Completed a master’s thesis on motion pattern prediction using learned spatial representations. Uber Technologies — Search Team San Francisco, CA ML Engineering Intern Jun 2023 – Aug 2023 - Developed XGBoost and DNN models for ETA prediction in driver-rider matching. - Built a dual-metric evaluation framework combining regression accuracy and Spearman rank correlation to optimize both low-latency inference and ranking quality. Stanford University — Hazy Research Lab Stanford, CA Researcher Advisor: Christopher Ré 2018 – 2019 - Extended Snorkel’s programmatic data labeling framework for multi-sentence relation extraction with Alex Ratner, now CEO of Snorkel AI. - Designed novel labeling functions and multi-task learning strategies, achieving a 12% F1 improvement via LSTMs. EDUCATION University of Illinois at Urbana-Champaign M.S. in Computer Science, 2023 GPA: 4.0/4.0 Fellowships: NSF GRF, Ford HM, GEM University of Illinois at Urbana-Champaign B.S. in Computer Science with Honors, 2019 GPA: 3.8/4.0 PUBLICATIONS Motion Pattern Prediction in Dynamic Environments Master’s Thesis, UIUC, 2023 Avoidance Critical Probabilistic Roadmaps for Motion Planning in Dynamic Environments F. F. Arias, B. Ichter, A. Faust, N. M. Amato IEEE ICRA 2021
Member since March 2026