Startup team in an enterprise
Download as PPTX, PDF0 likes168 views
This document discusses how SEEK Asia formed an autonomous start-up team to solve critical performance issues with their application processing. The team started by having engineers hack on a proof of concept to gain stakeholder buy-in. They then formed an engineering team to make the solution production-ready. The team deployed the solution and stabilized it before handing it over to other teams. Key traits of the start-up team included autonomy, an agile development lifecycle with frequent deployments, collaborative decision making, and a focus on monitoring, failing fast, and iterating enhancements. The recent example was a team that solved application indexing and search issues, allowing SEEK Asia to scale processing 10x faster while maintaining performance.
Startup team in an enterprise
- 2. Start-up Team in an Enterprise!How SEEK Asia embraces the culture of a Start-up to solve critical problems. Mazharul Anwar 2
- 3. Why do I need to listen to this guy? - I was with 5 start-ups ( ServiceClicks, txn.my, aquamob, tkt.my, CurrenSeek ) - Helped startups to design systems, build teams - Working for an enterprise for last two years. - Currently helping enterprise change and adopt to start-up culture.
- 4. 4 Who am I ? Mazharul Anwar Technical Lead, SEEK Asia @mazharul_anwar http://github.com/mazjs
- 5. When the team is formed ? - We have a critical problem to solve! - The problems that do not have right or one answer. - Example: Performance related issues, Build something to test out market fit etc. - Recent Issue : “ Scaling and performance issues with application processing. ”
- 6. Phases of the Team It starts with a hack. Hacking • A team of couple hackers start with a hack and builds a POC Buy In from Stakeholders • POC ( Proof of Concept) helps us to get buy in from the relevant stakeholders Forming an Engineering Team • Adding more engineers to the team to make this production ready. Deployment and Stabilisation • The team deploys and stabilise the product Hand-over to feature teams & Disband • Once stakeholders approves the product readiness ; team hands over the necessary documentation and knowledge. 6
- 7. Recent example of such a team in SEEK Asia We solved a single problem: Application Indexing & Search with ElasticSearch And some key results we achieved: - Able to scale 10x applications from current - Able to adapt changes fast - re-index < 3 days - Acceptable performance even scale to 10x
- 8. Traits of a Start-up Team in SEEK Asia Autonomous Team Development Lifecycle Feedback Loop Fail early & Fail often Monitoring Collaborative Decision Making; Sense of Ownership Enhancement Iterations
- 9. 9 Autonomous Team • Total autonomy in terms of : • Decision-making • Technology adaptation • System design and flexible delivery timings. • Traditional leadership roles became redundant because of the autonomy. • No Hierarchy
- 10. 10 Development Lifecycle • Extreme Programming • Micro Deployment Uber’s Micro deployment (https://eng.uber.com/micro-deploy/)
- 11. 11 Feedback Loop • Stand-ups • We were constantly on chat about design, problems, decisions with both team individuals and project stakeholders. • Helped us to do course correction and re-priorities (if needed) • We call it “Communication Driven Development” • Helps us to do cross team communication too.
- 12. 12 Fail early and Fail often • We failed! Many times over and some! • Giving up; or it wouldn’t work wasn’t in our vocabulary. We have to make it work. • We wanted to fail fast; so we could move on and try new ways of solving it. • We were not afraid to fail. • A guilty pleasure
- 13. 13 Monitoring • What gave us a huge boost in confidence to fail fast? Monitoring. • Building a monitoring system ( if there is none) is always the first priority. • Monitoring helps us understanding the problems and isolating them. • Micro deployment is not possible without monitoring. • We did shadow mode implementation of the system along with our monitoring. • Monitoring also helped us to implement a successful parallel system deployment strategy.
- 14. 14 Collaborative Decision Making • We failed as a team • Having a sense of ownership builds the confidence among engineers. • This along with autonomous team mentality give engineers the drive to build new things , try out new things in his/her own way. Sense of ownership
- 15. 15 Enhancement Iterations • Start with MVA and iterate • Sometimes this force us to do major re-architecting part of the system • Any time we could shed some ms ( milliseconds ) out of API calls, system calls or UI calls; we would do that. The best part is, this is done autonomously most of the time whenever any engineers can enhance any part of the system. • Understanding every part of the system by all the engineers enable us to do restructuring for better performance, re- architecting any part of the system easily as well if needed.
- 16. Q & A 16