PrathameshD_8 Profile Banner
Prathamesh Devadiga Profile
Prathamesh Devadiga

@PrathameshD_8

Followers
82
Following
61
Media
14
Statuses
91

Google SoC'25 @ UCSC | AI Research @ Dartmouth | Ex-Intern @ IIT- Indore | Member @ The Innovation Lab | AI Engineer | Founder & Lead Research @ Ādhāra AI Labs

Bangalore
Joined October 2022
Don't wanna be here? Send us removal request.
@PrathameshD_8
Prathamesh Devadiga
3 months
Hey everyone! I’m actively looking for AI/ML Engineer intern roles. Passionate about ML, DL, and building impactful AI solutions. If you know of any opportunities or connections, let’s chat! .Blogs: Github:
0
1
5
@PrathameshD_8
Prathamesh Devadiga
3 months
@diego_ongaro @JohnOusterhout 8/N: .Good sources to read about Raft in depth :. Original paper: Interactive visualization: Stanford lecture:
0
0
0
@PrathameshD_8
Prathamesh Devadiga
3 months
@diego_ongaro @JohnOusterhout 7/N: .6/ Raft adds safety through restrictions: Leaders can only append to logs, and votes are only granted to candidates with logs at least as complete as the voter's. In my Raft3D project, these principles ensure 3D printer data remains consistent even if servers fail.
1
0
0
@PrathameshD_8
Prathamesh Devadiga
3 months
@diego_ongaro @JohnOusterhout 6/N: .5/ Log replication is where Raft shines, the way it does is that it ensures all servers eventually store the same sequence of commands. The leader keeps retrying until followers successfully replicate entries.
1
0
1
@PrathameshD_8
Prathamesh Devadiga
3 months
@diego_ongaro @JohnOusterhout 5/N: .4/ Raft's way of handling failure: If a leader crashes, timeout mechanisms trigger a new election. Servers vote for candidates based on log completeness, ensuring the new leader has all committed entries.
1
0
1
@PrathameshD_8
Prathamesh Devadiga
3 months
@diego_ongaro @JohnOusterhout 4/N: .3/ How Raft works: A cluster elects a leader who handles all client requests. The leader replicates commands to followers, and a command is committed once a majority of servers have it in their logs.
1
0
1
@PrathameshD_8
Prathamesh Devadiga
3 months
@diego_ongaro @JohnOusterhout 3/N:.2/ Before Raft existed, we had Paxos which was imo brilliant, but notoriously difficult to understand. Raft broke consensus into clearer subproblems: leader election, log replication, and safety guarantees.
1
0
1
@PrathameshD_8
Prathamesh Devadiga
3 months
2/N: .1/ Raft is a consensus algorithm that helps distributed systems agree on shared state even when servers fail. Created by @Diego_Ongaro and @JohnOusterhout , it revolutionized distributed systems by prioritizing understandability.
1
0
1
@PrathameshD_8
Prathamesh Devadiga
3 months
1/N: .Fun weekend! .I built Raft3D: A distributed 3D printer management system that eliminates single points of failure through consensus!.Github Link: #DistributedSystems #Golang . But firstly, what's raft? Let me explain :.
1
0
4
@PrathameshD_8
Prathamesh Devadiga
3 months
Happened to work in Legal-AI space and was working with agents and fine tuning models. So, I fine-tuned Qwen2.5-7B w/ LoRA for Indian Law: JurisQwen. Will open source more as and when more additions would be done :D. @Alibaba_Qwen @huggingface.
0
0
2
@PrathameshD_8
Prathamesh Devadiga
4 months
Plan for the weekend? .- Just presented @ ICIAI, NTU Singapore :D.- Binge watch Andrej Karpathy's Deep Dive into LLMs video .- Read up on past blogs + finish up remaining paper writing work :D.
0
0
4
@PrathameshD_8
Prathamesh Devadiga
4 months
This is the 20th time I'm seeing this in span of 2 days :D
Tweet media one
0
0
0
@PrathameshD_8
Prathamesh Devadiga
4 months
Got selected to Google's GenAI Startup School Program :D Attended an insane set of classes so far!.@Google @GoogleStartups.
1
0
2
@PrathameshD_8
Prathamesh Devadiga
4 months
Just finished reading @MaartenGr and @JayAlammar 's Hands on Large Language Models Book. BEST BINGE READ EVER :D .On to @chipro 's AI Engineering book from tomorrow 🤩.(if anyone has other book recommendations, drop here!).
0
0
3
@PrathameshD_8
Prathamesh Devadiga
4 months
RT @aadhara_ailabs: Hello world :D.> I am an AI Lab based out of Bangalore, India.> here's my website : > We work o….
0
2
0
@PrathameshD_8
Prathamesh Devadiga
4 months
Literally did the same thing for my exam xD.and most of my midsem's mcqs were same as the questions asked during the exams!.
@mrnacknack
chirag
4 months
Midsems are easy af.> Run deepseek-r1 locally.> Feed all the study materials locally to the llm implementing RAG.> Collect all the PYQs, make deepseek to analyse it and answer descriptively according to the marks.> Tell it to give mermaid codes for flowcharts.You're gonna make it.
0
0
3
@PrathameshD_8
Prathamesh Devadiga
4 months
RT @yashwanthsai29: MCP is blowing up my timeline!.Seems like many still have questions about it. So, I’m starting a mini-series here on….
0
1
0
@PrathameshD_8
Prathamesh Devadiga
5 months
since, exams are done now, .started a smol ai org/lab to start building good research/open source ai projects. It's a small initiative, hoping to grow, I name it "Ādhāra AI Labs", Ādhāra - Foundational :D.Still a WIP website:
1
0
4
@PrathameshD_8
Prathamesh Devadiga
5 months
I think exam season makes me think about everything except exams because all new fun paper/project ideas to work on pops up my head now, SO much to do after this! (very excited :D).
1
0
4
@PrathameshD_8
Prathamesh Devadiga
5 months
RT @cneuralnetwork: I feel like you should never stop dreaming . my dream in 2024 was to get decent at any niche (which became ML).my dream….
0
23
0