Maxime Peabody
Austin, TX | maximepeabody@gmail.com
Current Role
Liftoff: Generative AI Engineer (May 2024 - present)
Technical lead on generative AI projects to improve efficiencies across different areas of the company, such as Sales, and ad creative generation.
Previous Experience
WOMBO.ai: Senior Software Engineer (Feb 2022 - April 2024)
Backend Engineer + ML ops/infra/exploration. Productionalized Machine learning models, Stable Diffusion (SDXL, controlnets, Loras), CLIP, BLIP, prompt engineering with GPT4, CI/CD, experiment framework, big data pipelines, AWS.
Google: L3, L4 SWE (2016 - 2022)
Backend engineer on Google Assistant, Search and Youtube Music. C++ and Java.
Detailed Experience
Liftoff - Generative AI EngineerMay 2024 - present
- Sales Agent: Designed and implemented an email sales agent driving over $10k monthly revenue. Tech: mailgun, postgres, GCP, docker, python, LLMs with instructor, weave.
- Slack Company Knowledge Assistant: Built RAG-based agent for company knowledge access using GCP, docker, python, Langsmith, pinecone, postgres.
- Automated Workflows: Saved 1000+ hours through LLM-powered automation of data analytics and email processing.
- Generative AI for ads: Fine-tuned image models using fal.ai and modal for custom ad creatives.
WOMBO.ai - Senior Software EngineerFebruary 2022 - April 2024
AI & Infrastructure:
- Led creation of generic ML Model interface, reducing productionalization time by 5x
- Implemented and explored SDXL, controlnets, text-to-video models
- Integrated GPT API for prompt improvement
- Reduced infrastructure costs by 10-25% through optimization
Google - Software EngineerSeptember 2016 - January 2022
Youtube Music (2019 - 2022):
- Improved recommendation algorithm for new music videos
- Implemented image generation pipeline with Skia
- Enhanced artist livestream notification metrics
Google Assistant (2017-2019):
- Implemented core Assistant features in C++
- Developed Android prototypes and improved web debugger
Side Projects
Socky.ai - CofounderMarch 2023 - Present
- Created sock designer tool enabling users to design custom socks using Stable Diffusion
- Built GCP infrastructure and API for business logic, AI server handling image generation, outpainting, and background removal using Stable Diffusion and Kubernetes
- Implemented GPT-4 for design suggestions and Gemini-Vision API for quality assessment
- Created automated pipeline for design collection generation and copyright risk assessment
Research Experience
Computational Geometry ResearchMay 2014 - August 2015
- Contributed to research on Local Routing in Convex Subdivisions
- Developed proofs for routing and broadcasting in planar subdivisions with low memory
Computer Vision ResearchMay 2013 - May 2014
- Researched computer vision techniques including image segmentation, edge detection and recognition
- Implemented algorithms using C++ and OpenCV: Canny Edge Detector, Hough Transform, Watershed Transform
Teaching Experience
Teaching Assistant - Introduction to Computers
Carleton University, Ottawa, Ontario (September 2013 - May 2014)
Provided support and guidance to students, focusing on fundamental computer science concepts including binary code
Education
Bachelor of Computer Science, Robotics Stream, Co-op Option
Carleton University, Ottawa, Ontario
GPA: 11.66/12
- Year abroad at the University of Edinburgh (2014/2015)
- Highest Faculty Entrance Scholarship (2012-2015)
- CodeGuru programming competition winner (2013, 2014)