Kevin Just
Data Scientist / AI Engineer
kevinjust.us/about-me
LinkedInData scientist, analyst, and AI engineer with 15 years of experience in tech. Growth roles at Facebook, Twitch, and TikTok, failed IPO at Wish, successful exit at LiveRail.
Product Analytics
Core metrics, funnel drivers, A/B testing, dashboards
Data Engineering
Pipelines, database design, efficient analysis foundations
AI Engineering
End-to-end software products with modern AI tools
Professional Experience
Marin Data Science·Data Scientist / AI Engineer·Fairfax, CA- Much of my work has been subcontracting with a digital strategy agency, working with ecommerce merchants in the beauty and fashion space
- Data pipelines connecting their Shopify (transactions, retention, LTV) data with their Google/Meta/TikTok Ads (costs, conversions, CAC) data for end-to-end ROI tracking and optimization
- Backoffice analytics suite: full funnel, retention, churn, and LTV reporting with breakdowns by network, campaign, creative
- Ad Campaign and Creative Management tools to streamline AdOps workflows
AskDatabot.ai /
LineupWiz.com·Data Scientist / AI Engineer (Personal Projects)·Fairfax, CA- askdatabot.ai: Full-stack AI platform — my goal was to build from scratch as much as possible - no templates, customized architecture
- Event tracking framework, backoffice analytics, Stripe billing, token-based subscription system, unit economics, and margin management
- Multi-model AI integrations: ChatGPT, Claude, Gemini, Mistral, Deepseek, Grok
- End-to-end traffic acquisition and marketing analytics pipelines with Google, Meta, and TikTok Ads
- LineupWiz.com: Fantasy baseball analytics platform leveraging the website architecture I've built
- Connects to Fantrax leagues, joins with MLB stat pipelines for unique analytics and roster management recommendations
ID.me·Principal Data Scientist·McLean, VA (Remote)- Background: ID.me has 130M users verifying identity for government services (tax filing, SSA, VA benefits)
- ID.me Shop offers discounts to verified Military, Nurses, Teachers, First Responders — my role was to scale this program
- Built automated reporting, optimization tooling, and customer segmentation models for targeted campaigns. Also hired and led a small team.
- Left to go all-in on using AI to enhance my skill set and build end-to-end products
TikTok·Lead Data Scientist, User Growth·San Jose, CA- Background: TikTok Shop was hugely popular in Asia, now launching in the West — shifting revenue mix from ads to commerce
- My role was to lead DS for User Growth: built team, led analytics/A/B testing/dashboarding for US product launch
- Ran experiments on coupon and discount strategies to drive first purchases
- Hired early UG team across Data Engineering, Data Analytics, and Data Science
- Worked with China team (proven success overseas) and "Project Texas" team (US data security compliance)
- Left due to RTO mandate, but learned a ton:
- Most sophisticated data warehousing I've seen — every metric expressed as multiplicative funnel decomposition
- E.g., Commerce $$ = Creators × Videos/Creator × Views/Video × Clicks/View × Add2Carts/Click × Purchases/Add2Cart × AOV × Margin
Wish·Staff Data Scientist, Pricing·San Francisco, CA (Remote)- Background: Hired to work on the 'Dynamic Pricer' — a legally questionable tool that maximized margins by optimizing price among user+product tuples
- Over time this was revealed to be essentially a price increaser to keep revenue high for IPO — the cost on user retention was EXTREME
- With a long-term holdout I found that this tool had actually cost the company more money over time than it gained, due to the impacts on retention
- What had once been the world's most downloaded app was on the way down
- Worked closely through a revolving door of 3 CEOs and 2 CTOs to safely do what we could to mitigate damage as the company lost over 99% of its value
Twitch (Amazon)·Senior Data Scientist, Ads·San Francisco, CA- Background: 2019, 5 years post-Amazon acquisition — the unit needed to stop operating at a loss, with the hypothesis that ads were a more scalable monetization strategy than creator subscriptions or bits
- Primary DS through many new product features: incentives for streamers to run ads, picture-in-picture viewing experience. Hired and helped build up the ads data science function from early stage
- Connected Amazon Ad Network to Twitch to programmatically fill and monetize unused inventory
- Ads revenue increased >50% Y/Y, attributed to these foundational ad tech improvements
Marin Data Science·Data Science Consultant·Fairfax, CA- Background: This was my first foray into independent consulting after a wild 3 years living through the Facebook acquisition and London move — and a chance to spend more time with my 2 young kids
- Most of my paid work came from latching on to a company that prototyped ideas for larger corporations. These connections led to similar work for other small tech startups
- Really fun and got me by, but I felt limited by the scope of work I could do at this point in my career
Facebook·Data Scientist·Menlo Park, CA / London, UK- Background: Sole data science carryover from the LiveRail acquisition
- Initially led a deep-dive measurement project quantifying the value of each unit of ad inventory the LiveRail Publisher Network was bringing to Facebook
- This analysis was the impetus for sunsetting LiveRail and rebranding it as Audience Network for Video with stricter quality standards
- Shortly after this we moved to London to ensure the successful launch of the Audience Network. I was the lead data scientist for this new product that we quickly grew to a $100M revenue run rate within 8 months.
- Extensive work on measuring and enforcing inventory quality — ensuring our inventory competed fairly in the auction vs Facebook/Instagram.
- I proved through brand engagement studies that our longer-form inventory deserved to use 10s view (vs 3s) as the conversion event
- This was pivotal for lifting off and reaching our revenue goals — our inventory was finally priced competitively for the value we brought to the massive FB ads ecosystem. We just needed to prove it objectively.
- Proposed and drove the sales incentive system for the first several hundred new publishers
- Bonuses correlated with audience value — this finally stopped the constant stream of low-quality, bot-driven, content-farm publishers that had been previously onboarded
- Facebook is where I learned a true model for Product Analytics: PM + DS + Design + SWE working as a pod toward the same goal. You could quantify the value of each product and builder — a key reason Facebook built so much, so fast. Gold standard that I carry with me.
LiveRail·Data Scientist·San Francisco, CA- Built tools for the LiveRail platform. My typical workflow was: create R package → pull data → process → output to database → dev team hooks into frontend
- Forecasting tool: helped publishers understand their total and unsold ad inventory
- Price floor recommendation engine: helped publishers maintain minimum prices for their inventory
- Real-time fraud detection: blocked shady networks from participating in realtime auctions
- Client reporting: Achieved 98% time and cost improvement in client reporting by identifying bottlenecks and optimal db design
Butler Shine Stern & Partners·Senior Data Analyst·Sausalito, CA- Led reporting for clients including Mini Cooper, Columbia Sportswear, and El Pollo Loco
- Built data pipelines using Hive on AWS EC2 to aggregate ad performance data to every Mini Cooper dealer in the country
- Inherited a contractor-built pipeline that took >24 hours and cost >$2,000/month in EC2 compute
- Rebuilt it using Python and R — cut runtime to <10 minutes and cost to <$10/month, a 99%+ efficiency improvement
- My first job after grad school, new to San Francisco — this is where I found my love for large systems and database design. First experience with git, command line, and AWS (much more complicated in those days). Made me excited to move into Data Science at a software company.
University of Michigan·Graduate Research Assistant·Ann Arbor, MI- Worked with grant recipients and academics on research projects as part of coursework and summer internships
- Summer Internship with an NSF grantee analyzing the effects of microtargeted ads on voter turnout and civic engagement. Very fun and interesting!
- Completed a study correlating modern Hip Hop music and Middle English iambic verse poetry. I data mined from patterns a grad student had codified across these genres.
- Taught 4 semesters of Stat 350 (Intro Stats) as a Graduate Student Instructor
University of Michigan ISR·Survey Specialist·Ann Arbor, MI- Worked on the NFL Retired Players Study — an early crack in NFL concussion management
- Contributed to the famous Survey of Consumer Attitudes
- Most of my time was spent conducting interviews and data entry — my first experience with applied data outside of my actuarial internships
- My ambition was to study Survey Methodology, but the opportunity to be a Graduate Student Instructor led me to Applied Statistics
Skills
Languages: TypeScript, JavaScript, Python, SQL, R, HTML/CSS
Frameworks: Next.js, React, Node.js, Tailwind CSS, pandas, scikit-learn
Tools: Git, Vercel, Supabase, PostgreSQL, AWS, Stripe
Domains: AI/ML Integration, Full-Stack Development, Data Analysis, Econometrics
Education
Master of Arts in Applied Statistics (Emphasis in Econometrics & Forecasting) - University of Michigan
- Most of my free time was spent practicing R programming and diving into internships and applied projects. For fantasy baseball, I transitioned from Excel spreadsheets to scripted modeling!
Certificate in Intro/Exploratory Statistics - Penn State
- Explored a career path in statistics
Bachelor of Science in Mathematics (Lower Division Honors, Actuarial Science) - University of Minnesota
- I was active on Math Team and competed in Men's Crew, traveling as far as Boston, Atlanta, Austin, and San Diego
- Passed Actuarial Exam 1 (Probability) and had a summer internship at Allianz Life Insurance Company in pursuit of an Actuarial Science career
K-12 - Montevideo Public Schools (Minnesota)
- Activities: Cross Country, Math Team (Captain, District Winner), Tennis, Knowledge Bowl — state team qualifiers in each
- Intel International Science & Engineering Fair (2000, Detroit)—"Analysis of Algorithms in an Iterated Prisoner's Dilemma"
- This was a unexpected and massively influential experience for a kid from a small farming community—probably contributed to me trying new and big things!
Way Back
Barnes & Noble·Bookseller·Minneapolis, MN- After Korea I went back to the University of Minnesota to take some Math classes as I debated whether to go back for a Master's degree
- Authors I got into this time around: Joseph Campbell, Carl Jung, James Joyce, mythology especially Native American
English Teacher·ESL Instructor·Masan, South Korea- 안녕하세요! I spent a year with my partner in a small town on the south coast of South Korea, teaching at daytime immersion schools and afterschool hagwons to students age 3-22
- We completely immersed ourselves in the culture and food — eating out almost every night, traveling almost every weekend
- We still eat kimchi every day, and I study Korean on Duolingo every night (강제로 — forced by my kid!)
Barnes & Noble·Bookseller·Santa Monica, CA- I moved down to Los Angeles to be an Actuary, but ended up having several small jobs including this stint at Barnes & Noble
- Possibly my favorite job of all time — I was the nighttime box unloader
- I started each night with ~35 big UPS boxes of books and 3 large, cleared tables surrounding me. I'd organize into ~20 piles which by midnight would be stacked over my head until I moved them onto carts for the shelvers in the morning
- I stumbled upon so many great books and authors I had never heard of, including Knut Hamsun and a book on the John Muir Trail — a 3-week backpacking trip from Yosemite to Mt. Whitney
- I ended up doing the John Muir Trail — one of the best decisions of my life!
Pizza Hut·Delivery Driver·Dinkytown, Minneapolis, MN- Legendary job driving around Dinkytown U of M campus delivering pizzas, making good tips for a college student
Just Machining·Machinist·Montevideo, MN- My dad started his own machine shop at 48 and ran it until retirement — a huge shift in his life and outlook
- Seeing his fulfillment left a big impression on me
- I worked for him after school in the offseasons, and most summers starting in 7th grade and through college
Star Tribune / West Central Tribune·Paper Carrier·Montevideo, MN- I woke up at 5:30am, 6 days a week, starting when I was 8 years old — how's that for work ethic!?
- I delivered ~40 papers a day and made $0.075–$0.115 per paper, about $100/month. I bought a 1991 Ford Taurus for $3100 when I was 16 with this money.
- I started each morning studying every baseball box score and the league leaders for the first 20 minutes!