Attended tastemaxxing masterclass at @waitin4agi_ 's office. Completely changed my perspective about content creation.
Also met @Papaocus and @techinmartin (Super cool guys)
Met @gkcs_ this Sunday
Some things that stuck:
1. Multi-agents sound cool but knowing when NOT to use them matters. This is still an evolving field.
2. LLMs still hallucinate.
20% basic work can be completely replaced and 60% can be partally done.
3. Tech meetups are worth it :)
@Hiteshdotcom 1/3 done β οΈ
Got placedd!!
Super thanks to @Hiteshdotcom and @piyushgarg_dev sir for GenAI Cohort and especially RAG is dead video!
Whole Interview was about RAG discussion.
The more I am going deep in deployment, more I am realizing why tools like jenkins exist and how much abstraction vercel's one click deployment really provide.
I've built a vey basic prototype using 330mb dataset using meilisearch and deployed it on my droplet.
Right now, it only does keyword matching and not semantic search.
Please do check out here:
https://t.co/6xnF2763wS
and also suggest how can I optimize it for large datasets :D
I've built a vey basic prototype using 330mb dataset using meilisearch and deployed it on my droplet.
Right now, it only does keyword matching and not semantic search.
Please do check out here:
https://t.co/6xnF2763wS
and also suggest how can I optimize it for large datasets :D
One of the most interesting projects that I did in college was to build a search engine from scratch on top of Wikipedia's data...
The best part was to create the entire search engine from the ground up without using Elasticsearch, Solr, or Lucene. I spent weeks building it and going through the nuances of information retrieval.
The dataset was 46GB, and parsing, processing, and storing the data on an i3 processor and 4GB of RAM taught me the importance of every single line of code we write. It also required me to understand how indexes are stored on the disk and their internal structures.
This is one project I recommend every single one of you to at least attempt. Apart from data management and processing, this project will also help you explore ranking algorithms and information retrieval techniques.
By the way, you can find the source code on my GitHub, and remember, this was done back in 2014 during my master's. So, coding practices are okay.
With the limited processing I had, I kept my laptop near the fan to avoid overheating for the whole night when I built the first index :)