The Day Three presentations of the 5th annual Python Web Conf are now up on LoudSwarm for re-watch. Here are some of the take-aways from each talk:
KEYNOTE: ClimateTech and the Future of the Planet MARK KOVSCEK @ CONSERVATION LABSKey points:
Mark presented a SWOT analysis for the planet Energy transition and technology advancements are reasons to be hopeful Threats to people include greenwashing, distraction, inaction, war, pandemics Strengths to rely on are human capacity and innovation Ultimately, most issues can be addressed by humans See how Conservation Labs reduces water waste and Co2 emissions KEYNOTE: Software Development in Distributed Teams DARVA SATCHER @ GITLAB, INCKey points = The values & behaviors Darva's global team relies on to be successful:
Work asynchronously and be explicit in your asks Document/record everything Be intentional with including everyone GitLab is sharing videos publicly about their best practices at https://www.youtube.com/c/GitLabUnfiltered
Django with PostgreSQL superpowers PAOLO MELCHIORRE @ 20TABKey points:
Postgres has the richest features of any database supported by Django. Postgres supports full-text search, array fields, fast random UUID updates on large database tables, indexes, aggregation function, range fields and more. Django add-ons for Postgres include very fast import of large tables, the ltree data type and range fields.
Make Your Python Web App Fly Around the World with CDNs MICHAEL KENNEDY @ TALK PYTHON TO MEKey points:
CDNs enable users to browse web apps superfast. Google Developers console only monitors web app’s performance for anonymous users. For logged-in users, Lighthouse comes handy. To simulate ping delay for testing, use Ngrok which will create a VPN tunnel to the local host running server. Check out Bunny.net (very fast and affordable). The Importance of Visuals in Teaching Python and Reducing Bias TADEH HAKOPIAN @ INDEPENDENT CONSULTANTKey points:
Making people comfortable with learning new things is key Removing barriers means: (i) make code non-intimidating; (ii) instead of text, provide visual representations and graphics; (iii) provide several methods of training; Interesting tools include PyFlow (https://wonderworks-software.github.io/PyFlow/ ) and Dynamo (https://dynamobim.org/ )
Modern typed python: dive into a mature ecosystem from web dev to machine learning SAMI JAGHOUAR @ JINA AIKey points:
Python's type hinting goes back over 10 years. Since then, type hints have become widely integrated. Static checking in the form of MyPy has been around, but things have accelerated with Pydantic — enabling frameworks such as FastAPI, SQLModel and Typer. DocArray (Sami’s project) is bringing the advantages type hints to ML projects.
Too Big for DAG Factories? CALVIN HENDRYX-PARKER @ SIX FEET UPKey points:
Airflow can scale, a good architecture of DAG files is key When working with DAGs make the import times very low Start with Dynamic DAG Factories, then statically generate to scale up Precompute configurations and data needed during build. Optimize to run tasks in parallel. TUTORIAL: Parlez-vous Wagtail? Internalization for Python/Django developers MEAGAN VOSS @ TORCHBOX, JACOB TOPP-MUGGLESTONE @ TORCHBOXKey points:
Learn how to create a Wagtail project Translate content into another language
App Security Does Not Need To Be Fun: Ignoring OWASP To Have A Terrible Time DWAYNE MCDANIEL @ GITGUARDIANKey points:
Being responsible for security can be isolating OWASP has resources like projects, community and best practices Check out the Top 10s (lists of vulnerabilities) Read the Cheat Sheet Series (for tech you use) Try out the Goats (deliberately insecure applications for training) Scan your app with ZAP (Zed Attack Proxy)
Celery on AWS ECS - art of background tasks & continuous deployment JAN GIACOMELLI @ RENKey points:
See recommended configuration settings for Fargate. Discover issues with deployment timing, working around long-running tasks Discuss strategies for tasks to facilitate more frequent deployment Explore workarounds for task locking The Five Stages of Leadership: The Journey from Coding to Leading RYAN BRUBAKER @ HAPPY DABBLEKey points:
Beware of hiring people similar to you Know your personality profile and the profiles of those around you Be a teacher for freshmen; supervise sophomores; coach juniors; delegate to seniors Invest in your people A few myths about GraphQL PATRICK ARMINIO @ APOLLO GRAPHQLKey points — GraphQL:
is not hard to learn is not used only with graph databases does not expose your whole database is secure can be cached does scale is not only for big companies 1-click deploys of Python web apps to Azure PAMELA FOX @ MICROSOFTKey points:
Deploying web apps can be tricky. There are several options to deploy to Azure, but you may not remember how to redeploy. Use Azure Bicep to deploy to Azure in IAC (infrastructure-as-code).
Come as You Are: Why Climate Tech Needs You and the Skills You Already Have JUSTIN MCCAMMON @ ENERGYHUBKey points:
The climate crisis is here, it just isn't evenly distributed yet. The solutions necessary to avoid catastrophic climate change are available today. People to make the solutions happen is what's missing. The skills you already have can be used to help solve the climate crisis. Every person has a place. Simplifying Real-Time ML Pipelines with Quix Streams: An Open Source Python Library for ML Engineers JAVIER BLANCO CORDERO @ QUIXKey points:
Streaming in ML scenarios can get you more real-time results. In an awesome interactive demo, Javier uses Quix Streams to connect browsers and Android phones in use by the audience to existing HuggingFace ML models. The app can detect sentiment analysis and shake detection in real-time, flagging chat messages as happy or sad and indicating who was shaking their phones. 
