Raj Chandrasekaran

Hi Adan,

Javascript has changed quite a bit in the recent years and lot of it was inspired from jquery. Now almost all modern browsers support javascript syntax everything that jquery does with few elaborate / sometimes better alternatives. So, if you like to switch, find the equivalents of what portions of jquery you use and replace those parts. Btw, jquery is still nicer sometimes with its method chaining and a lot simpler syntax - the equivalent in js may not be that sugary syntactically.

Have you explored ElasticSearch so far? You can build out simple PHP interface or use any of the CMS (wordpress) for your front-end

Hi Nicolas,

I think you need something like this: https://gist.github.com/imolorhe/f8794d3bb55e1a8065b23bcd0efeebe1 Using pdf.js you can convert each page in pdf to a corresponding image. I have not tested this myself, but PDF. js has been around for sometime and should work for you converting the pdf to image completely on the client side.

Hi Ashish,

If you need minimal work to be done from your end and like most of the components / design available out of the box - go with Bootstrap. This is the oldest and has the widest adoption and a whole range of components built out by others.

If you like Material design, this is a good choice too. Please note that Bootstrap also has a Material theme, though it is not as native.

Both of these above frameworks are bulky and has more than what you may need.

If you like to build micro-components in a elegant way, TailwindCSS is the way to go.

With the info you have shared so far, here's my recommendation: 1. Python ecosystem is very sophisticated for ML, AI. Build models, and perform most of background work via python. You can expose these models and any business logic needed via python Fastapi. 2. Node and js ecosystem is great for building both server rendered and Single page apps - whichever your preference is. You can easily use something like Next.js and build out the marketplace web/mobile application, interact with ML models via an api or directly read the model from node backend. Hope this helps.

Hi Akshay,

If the goal is to have a system that is purely to be used within your company, then it would be a good idea to have something scrappy. However, lot more has to be considered if you are planning to offer it as a SaaS or PaaS.

In either situations, the first factor I would consider is Return of Investment. You can build a really fabulous system, but if it is only going to help you little bit in the overall scheme of things, it may not be worth the effort. In other words, you can build - job management, scheduling, progress tracking, auto recovery, maintaining container state etc all by yourselves, but it may be worthwhile not reinventing the wheel when these solutions are already available (Jenkins or Team City for example).

If I were you, I will offload majority of the features like triggering jobs (build jobs), monitoring progress, errors etc to Jenkins. Every deployable unit can be dockerized, so that you get an uniform interface to validate whether the service is up and running.

You can build a thin vanilla UI on top of Jenkins' apis using React, or simply jQuery or Svelte. My personal preference would be Svelte.

If Jenkins is not a viable option, Go is the perfect backend for anything related to devops. You can use any Golang frameworks (Gin or Beego for instance) and have a front end in React / jQuery / Svelte. Hope this helps.

It purely depends on your app needs. Does it need to be scalable, do you have lots of features, OR it is a simple project with very simple needs - many of those parameters clarify which technologies will fit.

If you are looking for a quick solution, that reduces lot of development time, take a look at postgraphile (https://www.graphile.org/postgraphile/). You have to just define the schema and you get the entire graph-ql apis built for you and you can just focus on your frontend.

On frontend, React is good, but also need to remember that it is popular because it introduced one way data writes and in-built virtual dom + diffing to determine which dom to modify. Though personally I liked it, am recently more inclined to Svelte because its lightweightedness and absence of virtual dom and its simplicity compared to the huge ecosystem that React has surrounded itself with.

In all situations, frameworks keep changing over time. What is best today is not considered even good few years from now. What is important is to have the logic in a separate, clean manner void of too many framework related dependencies - that way you can switch one framework with another very easily.

Hi Sean,

Could you elaborate bit more? Is it that you are building a chat intermediation platform? OR just building a simpler proof of concept solution? Among the choices you are exploring, ScaleDrone and PubNub fall into one category. Firebase into completely another category - where you may have to build the chat features yourself, and Firebase just offers Realtime DB as a service (afaik).

If you are wanting to do a simple PoC, using existing chat applications and looking for hooks in that platform (be it ScaleDrone, PubNub, RocketChat (slack equivalent)) are few choices. Once you elaborate more, I can share some pointers based on what I may have come across.

Though we have always built something custom, Apache airflow (https://airflow.apache.org/) stood out as a key contender/alternative when it comes to open sources. On the commercial offering, Amazon Redshift combined with Amazon Kinesis (for complex manipulations) is great for BI, though Redshift as such is expensive.

React End to End or even Meteor JS will work for your needs if you are planning to build it yourselves. The framework selection depends on volume, usecases, desired user-experience and most importantly Return of Investment (ROI). Please elaborate what usecases are you trying to solve for.