How RnD Appears in Large Tech Companies
Google · Amazon · Yandex · Tinkoff
Slide contents
1. How RnD Appears in Large Tech Companies
Google · Amazon · Yandex · Tinkoff
2. This Is an Author Model, Not an Insider View
Publicly available information only
Every company uses the same comparison frame
Conclusions reflect the author's interpretation
3. The Same Questions for Four Companies
How the business grew
What sustains the business model
How the research philosophy is framed
Which systems, papers, and products emerged
4. 01. Google
A research portfolio and papers that became industry infrastructure
5. Search Became an Ecosystem of Touchpoints
6. User Intent Connects Both Sides of the Market
7. Research Is a Portfolio of Risks and Horizons
Fundamental questions and new capabilities
New products and lines of business
Research contributions to active products
Infrastructure that multiplies team impact
8. One Project Can Span Multiple Quadrants
9. Value Emerges at the Boundaries Between Types
10. Internal Infrastructure Gave the Industry a Vocabulary
11. First a Unique Internal System, Then an Industry Analog
Complex systems were built and documented first
Open analogs diverged from the internal stack
Incompatibility limited community help
Exceptions: Android, Chrome, Kubernetes, TensorFlow
A paper spreads the idea; compatible code spreads the practice
12. 02. Amazon
Customer obsession and engineering that became a public cloud
13. A Store Grew into a Marketplace, Media, and Infrastructure
14. Prime and FBA Reinforce Both Sides of the Marketplace
15. Research Starts with Customer Pain
16. The Problem Defines the Research Area, Not the Org Chart
17. Before 2015, the Output Was Usually a Service, Not a Paper
18. First the Product and Cloud Service, Then the Publication Stream
Engineering results became products early
There were comparatively few papers before 2015
AWS turned internal practices into a market
S3, Availability Zones, and Regions became de facto standards
A standard spreads through a working service
19. 03. Yandex
Homegrown infrastructure and an accelerating open-source turn
20. Search and Ads Repeat Google's Two-Sided Logic
21. A Homegrown Stack Comes from Six Forces, Not One
22. In Two Years, Open Source Became a Portfolio, Not an Exception
23. Leadership in Complex Systems Stayed Internal for a Long Time
Many complex solutions were built among the first
Open analogs diverged from internal tools
That limited the value of community help
Opening systems accelerated visibly from 2022
It is time to rebalance internal strength and external compatibility
24. 04. Tinkoff
A hybrid RnD-center proposal for long-horizon engineering problems
25. An Online Bank Grew into an Ecosystem for 38 Million Customers
26. A Hybrid Center Joins Academic Depth with Business Scale
27. A Paper and a Prototype Are Transitions, Not Endpoints
28. Research Themes Grow from Engineering Constraints
29. Research Assets Were Already Becoming Products
30. Similar Problems Spread Through Different Vehicles
31. RnD Appears Where the Product Horizon Becomes Too Short
The business model sets the research direction
Scale and unique constraints require proprietary knowledge
Papers, services, and open source spread knowledge differently
There is no single correct organization model
RnD is the bridge from long-horizon uncertainty to repeatable business impact
32. What the Comparison Is Based On
research.google/philosophy
amazon.science/about
Why big tech infrastructure usually consists of home-written solutions
Public product timelines, papers, and company open-source releases
33. Thank you!
Knizhny kub
Continue the discussion: t.me/book_cube/1799
Alexander Polomodov, Technical Director & Fellow, T-Technologies
@book_cube