$ independent AI / ML contractor

Practical AI & ML
for real business problems.

I'm Vadim von Brzeski, a data scientist and AI leader with 20+ years in machine learning. I help companies apply AI, ML and modeling where it makes a difference, either by owning a project end to end or by joining your team as a senior, hands-on contributor.

Previously: Director of Marketing Data Science, eBay · Yahoo! Search & Yahoo Labs

Three ways I can help

λ₁

Solve a specific problem

You have a business problem and suspect AI or ML can help. I'll scope it, build the solution, and hand it off in a form your team can run.

  • Figure out whether AI/ML is the right tool
  • Data exploration and feasibility check
  • Model or AI system built and validated
  • Deployed, documented and handed off
λ₂

Join your team

You already have a team and need more senior AI/ML capacity. I plug in as a contractor, work in your stack and process, and start contributing quickly.

  • Hands-on modeling and engineering
  • Technical lead or second opinion on approach
  • Mentoring and code / model reviews
  • Short-term or ongoing engagements
λ₃

Lead your data science effort

You need senior data science leadership but not a full-time director. I step in part-time to set direction and help your team deliver.

  • Priorities and roadmaps, set with product leaders
  • Technical and architecture direction
  • Mentoring and coaching for data scientists
  • Partnering with engineering to get models into production

What I work on

  • LLM fine-tuning
  • Generative AI for marketing & content
  • Reinforcement learning
  • Predictive & propensity modeling
  • Churn & customer behavior modeling
  • Recommendations & ranking
  • Causal inference & impact measurement
  • End-to-end ML: data to deployment
  • Data science leadership & mentoring

Industries

  • E-commerce
  • Marketing
  • Search & advertising

Experience you don't have to manage

20+ years in machine learning & AI

Most recently I was Director of Marketing Data Science at eBay Core AI, where I led scientists and engineers applying AI to e-commerce and marketing. Before that, I was an applied researcher at Yahoo! building machine-learned systems for search and advertising. And before that, I was the CTO/Founder of a dot-com era startup developing cloud-computing solutions.

I've been building models since long before "AI" was a buzzword. That long view helps me separate what will actually work for your business from what just demos well.

LogLambda Consulting LLC is my independent practice. When you work with LogLambda, you work directly with me. No account managers and no junior staff.

See my background on LinkedIn →

01 Business problem first

I start with what you're trying to achieve, not the model. If AI isn't the right answer, I'll tell you.

02 Prove it early

I validate ideas on your real data before you commit to a bigger build.

03 Measure what matters

Results are tied to metrics you care about, so you can see whether it's working.

04 Fit in, hand off cleanly

I work the way your team works and leave clean code, clear docs and no dependency on me.

Education

  • M.S., Statistics & Applied MathematicsUC Santa Cruz
  • M.S., Computer ScienceUC Santa Cruz
  • B.S., Electrical Engineering & Computer ScienceUC Berkeley

Problems I've worked on

A sample of production ML/AI projects from my time at eBay and Yahoo!.

eBay · GenAI

Personalized marketing copy

Fine-tuned open-source and proprietary LLMs to generate personalized marketing copy for campaigns.

  • LLM fine-tuning
  • Open-source & proprietary LLMs

eBay · GenAI

Listing title & description rewriting

Fine-tuned LLMs to rewrite item listing titles and descriptions on ebay.com to drive buyer engagement.

  • LLM fine-tuning
  • Content generation

eBay · GenAI

Shopping intent detection

Developed LLM-based solutions to detect users' shopping intent at scale.

  • LLMs
  • Intent detection

eBay

Marketing communications optimization

Optimized the frequency and channel (email vs. push) of marketing communications to lift engagement without increasing unsubscribes.

  • Reinforcement learning
  • Email & push

eBay

Purchase intent & propensity

Inferred shoppers' purchase propensity from their activity to drive promotional marketing emails and product recommendations.

  • Spark MLlib
  • GBDT
  • PyTorch

eBay

Churn & re-activation

Predicted which inactive users were likely to come back, so marketing could target re-engagement where it would pay off.

  • Spark
  • MLlib

eBay

Measuring the impact of app releases

Estimated the causal effect of new app releases on user behavior, using large-scale longitudinal modeling of user activity.

  • Causal inference
  • C++

Yahoo! Search

Entity detection & ranking

Science and technical lead on a platform that detected key phrases and named entities in news, mail, finance and sports content, with ML ranking, summarization and word-sense disambiguation.

  • NLP
  • SVM
  • GBDT
  • LDA

Yahoo Labs

Click prediction for search ads

P(click) models for sponsored search ads, including large-scale data exploration, feature engineering and model verification.

  • Hadoop
  • Pig
  • Python

Have a problem worth solving?

Or a team that could use another experienced hand? Send me a short note about what you're working on and I'll get back to you.