A
Alex Rowan
About
Detail
Houston, Texas, United States
Résumé
Jobs
verified_user
0% verified
- VSenior AI EngineerVellum AIJan 2024 - Current (2 years 9 months)• Fine-tuned LLMs for multiple industries, increasing domain-specific response
accuracy by 30% for enterprise clients.
• Built agentic RAG pipelines using LangChain/LangGraph, reducing incorrect
responses by 25%.
• Optimized on-device models (quantization & distillation), lowering inference latency
by 40% on mobile and edge devices.
• Developed real-time voice AI prototypes supporting 500+ concurrent users for end-
to-end speech → reasoning → speech workflows.
• Systematised ML workflows with MLflow, Docker, and CI/CD, cutting deployment
time by 50%. 
AI EngineerEleven LabsMay 2022 - Dec 2023 (1 year 8 months)• Built predictive and generative AI models achieving 85%+ accuracy, improving sales
and marketing forecast reliability.
• Mechanized ETL and annotation workflows, reducing manual processing by 40% (-15
hours/week saved).
• Enhanced internal dashboards enhancing decision-making speed by 20% across 10+
teams.
• Maintained ML pipelines handling 50,000+ samples/month, ensuring robust
training, validation, and deployment.- AI Software EngineerBasetenOct 2019 - Apr 2022 (2 years 7 months)• Improved LLM-based AI solutions improving enterprise predictive analytics accuracy
by 20%.
• Built deep learning models for time-series and spatio-temporal data, achieving 92%
prediction accuracy on key datasets.
• Automated data pipelines processing 100,000+ records/month, reducing manual
errors by 30%.
• Collaborated on AI optimization projects, improving inference speed by 35% for
production systems.
Education
verified_user
0% verified
- Machine LearningDeeplearning.ai
- Data Science and Artificial IntelligenceBoston Institute of Analytics
- Deep Learning SpecializationDeeplearning.ai
- BBachelor of Science in Computer Science
- TensorFlow DeveloperDeeplearning.ai
Projects (professional or personal)
verified_user
0% verified
- DDeepFake Audio Detection• Executed Sense Voice for feature extraction and AASIST for classification.
• Achieved 98.89% accuracy on ASVSpoof2021 dataset; deployed as FastAPI service with Docker. - RRAG-based AI Doctor Chatbot• Fine-tuned Llama-31B on ~2,000 medical QA samples, improving medical answer accuracy by 28% and reducing
hallucinations by 35% in evaluation tests.
• Built agentic RAG pipelines using LangChain, reducing incorrect responses by 25% and improving retrieval
precision across 50k+ documents.
• Deployed on AWS (Lambda + API Gateway) with MLflow for versioning and metrics. - LLLM-Powered Marketing AI Assistant• Designed a multilingual real-time voice bot using STT → LLM → TTS pipeline.
• Implemented streaming audio via WebRTC for live audio processing.
• Packaged as a FastAPI service with modular transcription, reasoning, and audio generation components.