Jason Liu¶
ML Engineer, Consultant, Advisor
[email protected] | jxnl.co | jxnl | jxnlco
Executive Summary¶
I'm a former Staff MLE who built a 7-figure AI business by bridging the gap between research, product, and customers. I created Instructor (3M+ monthly downloads) and have spent 10+ years building AI systems at scale while training 1000+ practitioners. I help companies establish thought leadership in agent engineering, build evaluation frameworks from customer insights, and drive developer adoption at scale. I'm looking for my next 4-5 year project to apply this expertise.
Professional Experience¶
Founder, 567 Studio¶
2023 - Present
- Work with Seed to Series B companies, helping engineering teams adopt AI best practices and upgrade existing LLM systems to agentic applications across Retrieval Augmented Generation (RAG), Context Engineering, and Evals.
- AI consulting for Distributional, Galileo, Zapier, HubSpot, Limitless, Tensorlake, Trunk Tools, New Computer, Weights & Biases, Modal Labs, Timescale, and others across construction, medical, CRM, and personal assistant industries
- Built training programs on improving AI systems on Maven serving professionals from OpenAI, Anthropic, Google, Microsoft, Amazon, Netflix
- Designed workflow automations, fine-tuned embedding models for images/diagrams/videos across industries (consulting, oil & gas, meeting notes, red teaming, medical, crm, personal assistants)
- Conducted AI adoption training for enterprise CTOs and engineering teams, establishing monthly lunch-and-learn programs and best practices frameworks
Staff Machine Learning Engineer, Stitch Fix¶
2018 - 2023
- Architected multimodal embedding similarity system using ResNet-50 transfer learning and CLIP+GPT-3 integration generating $50M+ annual revenue impact
- Designed Flight framework handling 350M+ daily requests with 80% internal adoption rate across engineering teams
- Built styling chatbots using DaVinci chat models, integrating LLMs for stylist workflows and improving recommendation quality by 40%
- Technical lead for team of 6-7 engineers and data scientists
Side Projects¶
- Instructor Python library (4.5M+ monthly downloads), recognized and referenced by OpenAI. Technical documentation, API design, and developer onboarding experiences.
- Data Science Arena system for training AI judges on analysis quality, enabling preference collection and evaluation beyond code correctness
Data Scientist, Facebook (Meta) - Protect & Care¶
2017
- Developed ML algorithms detecting harmful content (violence, hate speech, child endangerment) with 95%+ accuracy at 2B+ user scale
- Built real-time monitoring dashboards identifying malicious activities, reducing escalation time by 50%
- Contributed to platform integrity improvements protecting global user base
Education¶
Bachelor of Mathematics in Computational Mathematics & Statistics
University of Waterloo, Ontario, Canada | 2012-2017
Background: Mathematical physics and computational statistics before transitioning to applied machine learning
Publications & Research¶
Liu, J., Weitzman, E.R., & Chunara, R. (2017). Assessing behavior stage progression from social media data. Proceedings of the 2017 ACM Conference on Computer Supported Cooperative Work and Social Computing, 1320–1333.
Rehman, N., Liu, J., & Chunara, R. (2016). Propensity score matching for vaccination sentiment analysis. AAAI Spring Symposium on Observational Studies through Social Media and Other Human-Generated Content, 23–25.