T-Meetup: R&D in Vladivostok (RnD heading)
T-Meetup: R&D in Vladivostok (Rubric #RnD)
I spent most of this week with colleagues from the RnD center in our development center in Vladivostok. (It is very nice and delicious food here:)). And tonight, I've been there. RnD mitapThe program of which included three reports on the differences between RnD activity and mere development and a couple of specific cases of real research.
- “Why are there problems that cannot be solved by conventional development?” – Stanislav Moiseev, Director of Engineering Research at T-Bank (We flew together from Moscow.)
- “Code Knowledge Graphs as a Memory for LLM” by Mikhail Baderik, Rnd Engineer from our center in Vladivostok
- “Evolutionary Agents in Practice: Open Source and Real Tasks” by Danial Matienko, Rnd Engineer from our Vladivostok Center
I wanted to talk about Stas’s report separately, since he basically talked about the concept of research and development in large IT companies, in particular, in T-Bank.
Stas showed how RnD differs from product development:
- Product development answers the question “what to do and how to do”, using ready-made “instructions” R&D answers the question “can this be done in principle?” – it is about closing technological risks by testing hypotheses and developing new methods and technologies.
Different countries approach RnD differently - USAA model based on large capital. Includes startups, corporate research centers (Apple closed, Microsoft Research fundamental, Google hybrid) and strong university labs. By the way, at the end 2023 year stripper Google approach in podcast (Stas also participated as a guest.) The hybrid approach is that they deeply integrate research teams into product processes. Their key principle is that the reseasrch team writes production code from day one to avoid the separation of research from practice and accelerate implementation. **- China.**A model based on cooperation and state influence. It is characterized by close cooperation between companies and universities, centralized planning and significant public investment. (Example: Huawei is spending more 20Percentage of R&D revenue). - Russia.: Historically strong model with industry research institutes. Today, research is increasingly moving inside large IT companies. The need for import substitution and the rapid development of AI stimulate the growth of investment in R&D.
Next Stas told about the direction of engineering RnD in T-Bank 1. Engineering productivity: Develop AI-based tools to improve the efficiency of engineers. Examples: CodeReview service, unit test generator, Data Scout tool for automatic data collection. 2. Large graph analysis: Create your own platform for processing graphs that do not fit into RAM. This is necessary for solving problems in the field of recommendations, antifraud and infrastructure analysis. 3. Optimizing the data platformDevelopment of methods for accelerating SQL requests due to approximated calculations on sampled data, which allows achieving 10- multiple acceleration while maintaining 99Percent accuracy. 4. Optimization of logistics: Application of hierarchical clustering algorithms to optimize courier routes. Implementation of 16 The region has already reduced the mileage 9%. 5. Blockchain: Direction related to transaction analysis, fraud detection and creation of investment instruments based on distributed ledgers.
Stas concluded his story by saying that R&D teams need not only a budget and challenges, but also a reasonable time frame, a talented team and a culture that gives researchers a sufficient degree of freedom.
Well, if you whip up useful ideas from the report that you can take yourself, then they are something like this. 1. Assess technological risks when launching new products 2. Integrate research and development (Avoid creating detached research teams) 3. Study Open Publications 4. Look for opportunities to optimize 5. Create the Right Culture
#RnD #Engineering #Software #Management #Leadership