Project Clarity
~5 mins read
QA
current pains vision past learnings good bad ugly
- Past, present, future for the company, team, person, and lifecycle
- historical context
- Present challenges and opportunities
- Future vision and direction
- Lifecycle
- planning,
- dev,
- testing,,
- deployments,
- monitoring,
- on-calls what could be better about the life-cycle? what do you like about it?
- team dynamics
- which teams are there
- how do they collaborate
- who is on this team and for how long
- Good bad ugly
- things to improve
- things they like
- churn/burn out/hardest parts
- Growth
- what have they been learning
- how the work improved their skills (look if they just focus on tech)
courses
- sociology class
- humanities
- art of animation
- music
- advanced swimming
-
econ
- accounting
- probability
-
deutsch A2
- entrepreneurship
- IoT
- startup class
-
tunga gungor, algorithms
- satellite space
-
info retrieval, sentimental analysis project
- CNN lung disease detection
- electromagnetic field theory
- robotics
- microcontrollers, doodle jump, overcomplicated trajectory
- pattern recognition
- modern physics
planned Senior project poorly and did most of the work on the last week
overall, I overloaded myself and caused a lot of unnecessary stress, feel lucky to not get severe burnout
context
real estate analytics for an alumni
yacht market analytics
cosmetics analytics
retail analytics, supermarket products
- Crawl
- Set up a data pipeline
- Implement the business logic
- Sync to ElasticSearch and Metabase
- Create an API
- Create an orchestrator
- Set up monitoring
- Create recovery procedure
- Security review
entropy, manual decision tree > stats > NLP
then tried to create a generic system for all this, but itโs harder than it sounds, also given my overall experience at the time
- negotiation and pricing mistakes
- underestimating the cost to me and the benefit to the customer
- clarification mistakes,
- should only put enough effort to verify an approach
- distractions of a small business, accounting, office space, bills, contracts, โฆ
-
slow reporting, measure & debug, found the gitlab was the bottleneck, replaced with gitea
-
manager change, guidance through identifying and prioritizing, charting the technical roadmap and communicating through both sessions and docs
-
driven monitoring, log collection, telemetry, DevOps collab
-
error codes feedback
- inter-team collaboration feedback
- visibility/communication feedback
DP
problems
- minio overeng story
- redis debug story
- reinventing dbt
- duct tape
- a custom system, hard for data engs
-
slow builds, rewrite dockerfile, remove deps, parallelize tests and CI steps
- reverse engineered and diagrammed
- driven the discussion and planning, even discussed with the director and the CTO
- laid out the problems and our possible actions with their tradeoffs
- created an MVP and iterated it to feature parity and beyond
- simple to complex, standard library vs an orchestrator dependency, redis to postgres
- then led people to integrate it to other systems
- got 4 swes and 3 data engineers up to speed
-
and had the happiness to see it flourishing
- new airo, adding people wonโt help with deadlines
- onboarded juniors, no prisma orm, plain postgres
- dependency script
blackops
concept to production in 6 weeks manual entry bug initialization bug currency conversion bug, 10x the size, exposing us to more risk 5-7 concurrent traders, $20k, $1m trades per day
MS
MS onboarding volunteered for accessibility quickly joined on call for a complex system with 2 stacks and > 5 teams
WIP
FEEDBACK
fb: error codes
approach it as tech discussion of numbers vs names as enum keys rather than criticism
we decided to change the approach
fb: distractions after a manager change
shields vs funnels talk with my manager
havenโt solved but helped
DIFFICULT SITUATIONS
dif: minio overeng
suggested a session to address their concerns
led to improvements about deployment and monitoring processes
dif: nm, focus on one vs many
delegated but not really worked
dif: pushback against a deadline
a major release in 3 months and CTO meeting, shift resources
having new people are great in the long term but wonโt help with releasing earlier in the short term
he already knew this but such feedback from the team helped to clarify expectations
COLLABORATION
guided an unhappy teammate in a chaotic period
provided direction with suggestions for things we can improve, pair programming
he turned out to be one of the most effective people in the team
onboarding 3 junior engs
created onboarding docs
creating helpful starting tasks
pair programming
no data engs
cover but advocate
doodle jump
overcomplicated trajectory
process the main character image
TECH
tech: slow gitlab
tech: driven monitoring work
little structure, no telemetry
better monitoring
##
stories are about people
so if I think of people I worked with, I can come up with some interesting cases, and the stories donโt have to be extraordinary