| Author | Message | ||
Baatasari Comedian Username: Baatasari Post Number: 1098 Registered: 10-2018 Posted From: 171.159.64.10 Rating: N/A |
i think thats where.. you need to be very CLEAR .. these 2 profiles are generally put in a very Grey Area like 'data scientist' 'data engineer' .. any enterprise deployment, as you see, you need BOTH the people.. at times both are same, but to my knowledge, people are differentiated.. Sagemaker meeku ready-made ichedi 1) matrame => danni bare minimum use cheyalanna, you need to know ML... and most cases you need to then put your own toppings on the offering.. ika 2) is someone who is deploying, he NEED NOT know any ML.. veediki Hadopp,spark, etc ee typical Tech profile unte chalu, he can treat the ML as blackbox and still move on with job.. I see typical IT folks starting as 2) trying to encroach slowly into 1).. i guess you are also someone trying 2) => 1) brahma satyam jagan-mithyä jivo brahmaiva näparah | ||
Emc2 Legend Username: Emc2 Post Number: 72499 Registered: 03-2008 Posted From: 216.81.94.71 Rating: N/A |
unnayi ,we are using it , but ML and AI has different architecture in terms of scalability It's ok | ||
Emc2 Legend Username: Emc2 Post Number: 72498 Registered: 03-2008 Posted From: 216.81.94.71 Rating: N/A |
this is my interested area, but isn't it sagemaker does the same job, of course with more infra features together anuko. It's ok | ||
Biggboss Side Hero Username: Biggboss Post Number: 7268 Registered: 08-2017 Posted From: 45.62.177.6 Rating: N/A |
BRMS based inka unnayi...good alternative leka, people are still using it | ||
Baatasari Comedian Username: Baatasari Post Number: 1097 Registered: 10-2018 Posted From: 171.159.64.10 Rating: N/A |
I am a fan of the AI/ML ecosystem developed by Amazon & MSFT.. more fanism with Amazon .. ee topic Madhwapathi we discussed in the DB.. it ultimately boils down to 2 profiles, depending on your work - Inner, Development work => you write the ML algos, so for you be it AWS or MSFT they give you plug&play features.. though its a bit restricting if you try to develop exotic custom models Outer Deployment => more of infra job, deploying, supporting, scaling the product AWS suite of products, i feel are here to stay.. and for enterprise if you are from 1) i suggest you concentrate more on the ML part, less on Sagemaker.. if you are 2) you need to be taking sides - AWS gang or Azure gang.. master one atleast.. MP could add more details.. brahma satyam jagan-mithyä jivo brahmaiva näparah | ||
Emc2 Legend Username: Emc2 Post Number: 72497 Registered: 03-2008 Posted From: 216.81.94.71 Rating: N/A |
this is exactly i am also thinking about..i used to do this using BRMS products like ilog and drools but i feel those are old fashioned.. It's ok | ||
Biggboss Side Hero Username: Biggboss Post Number: 7267 Registered: 08-2017 Posted From: 45.62.177.6 Rating: N/A |
Ad Tech (Retail vertical) Use case vachi - serving tailored ads based on product recommendations | ||
Emc2 Legend Username: Emc2 Post Number: 72496 Registered: 03-2008 Posted From: 216.81.94.71 Rating: N/A |
what industry you are in? business case enti? It's ok | ||
Biggboss Side Hero Username: Biggboss Post Number: 7266 Registered: 08-2017 Posted From: 45.62.177.6 Rating: N/A |
I can give you one use case there are lot of Shopify, BigCommerce clients who do not have infra to build their own product recommendations like Amazon's similar product recommendations not a lot of players offer this feature as a third party plugin think about it.. | ||
Biggboss Side Hero Username: Biggboss Post Number: 7265 Registered: 08-2017 Posted From: 45.62.177.6 Rating: N/A |
if you are manually sacling it, chaala painful we often see memory skew with our data (not distributed uniformly), alanti cases manual ga tune chesthamu parameters ...and each run takes 3 hours to validate thala noppi...ee big data frameworks tho | ||
Emc2 Legend Username: Emc2 Post Number: 72495 Registered: 03-2008 Posted From: 216.81.94.71 Rating: N/A |
use case emi ledu, exploring and experimenting as my pet project but ninna ma komfany lo free training work shop unde, i missed it, komfany email chuda ledu trying to find who is using it and where can we use it, also looking to build as a product like any one can customize it anattu It's ok | ||
Enigmatic Comedian Username: Enigmatic Post Number: 1912 Registered: 11-2010 Posted From: 69.249.71.186 Rating: N/A |
aws emr cluster meeda spark ml bane scale avutundi anukunta. | ||
Biggboss Side Hero Username: Biggboss Post Number: 7263 Registered: 08-2017 Posted From: 45.62.177.6 Rating: N/A |
what's your use case? what kind of recommendations are you planning to generate? | ||
Emc2 Legend Username: Emc2 Post Number: 72494 Registered: 03-2008 Posted From: 166.137.244.125 Rating: N/A |
Just copy ante kastam le kaani, I saw those algos they are like our math theorems They will have their own implementation of it plus it is not just algorithms you need to scale it too It's ok | ||
Biggboss Side Hero Username: Biggboss Post Number: 7262 Registered: 08-2017 Posted From: 45.62.177.6 Rating: N/A |
https://ttvand.github.io/Second-place-in-the-Santander-produ ct-Recommendation-Kaggle-competition/ | ||
Biggboss Side Hero Username: Biggboss Post Number: 7261 Registered: 08-2017 Posted From: 45.62.177.6 Rating: N/A |
they don't...kani nenu vinnandhi enti antey--they pretty much copy open source code and not give credit or contribute back | ||
Enigmatic Comedian Username: Enigmatic Post Number: 1911 Registered: 11-2010 Posted From: 69.249.71.186 Rating: N/A |
does amazon provide their implementation of general machine learning algo's in their platform? | ||
Biggboss Side Hero Username: Biggboss Post Number: 7260 Registered: 08-2017 Posted From: 45.62.177.6 Rating: N/A |
one this for sure,AWS Sagemaker is very expensive Our internal system generates 600 million product recommendations every day with minimal cost (approximately 10k USD for these services- 30 node EMR cluster, Dynamo DB, S3) | ||
Biggboss Side Hero Username: Biggboss Post Number: 7259 Registered: 08-2017 Posted From: 45.62.177.6 Rating: N/A |
I am exploring it, we are planning to replace our product recommendations system with Sagemaker so far no success, next quarter kalla finish cheyyali before we switch to this system, we need to spend lot of time doing A/B to measure sagemaker efficienty compared to our own internal recommendations system | ||
Emc2 Legend Username: Emc2 Post Number: 72493 Registered: 03-2008 Posted From: 216.81.94.71 Rating: N/A |
ninna demo and training unde.. i missed it.. evarikina exp unda deentlo aws AI and machine learning? what are the prospects who is using it? It's ok |