_AbhinavJ Profile Banner
Abhinav Profile
Abhinav

@_AbhinavJ

Followers
143
Following
7K
Media
17
Statuses
127

CS @uwaterloo

Joined November 2023
Don't wanna be here? Send us removal request.
@_AbhinavJ
Abhinav
3 months
demo btw .
0
0
5
@_AbhinavJ
Abhinav
3 months
@cohere @tavilyai @JinaAI_ "custom deep research" . @1vnzh, @aidangomez
Tweet media one
1
0
6
@grok
Grok
2 days
Join millions who have switched to Grok.
122
238
2K
@_AbhinavJ
Abhinav
3 months
AccessPath. using AI to analyze and make physical spaces more accessible. built a custom deep research pipeline using @cohere, @tavilyai, @JinaAI_
Tweet media one
Tweet media two
Tweet media three
2
1
15
@_AbhinavJ
Abhinav
5 months
MCP Server for @toggl. Track your time efficiently.
1
0
9
@_AbhinavJ
Abhinav
5 months
@Google @GoogleCloudTech @MichaelVakoc I remember seeing you guys working in the agent space so would love to hear what you guys have to say too!.@himanshustwts @doesdatmaksense @DhravyaShah @gregisenberg.
0
0
1
@_AbhinavJ
Abhinav
5 months
@Google @GoogleCloudTech @MichaelVakoc I'd love any suggestions, clarifications, corrections, anything of the sort!.
1
0
1
@_AbhinavJ
Abhinav
5 months
travelled 8+ hours for the @SocraticaInfo Symposium to demo autonomous swarm intelligence. on the way back, won the best use of @Cohere at @GenAIGenesis, the largest AI Hackathon in Canada, .after soloing the entire ml pipeline in 24 hours. @1vnzh
Tweet media one
Tweet media two
17
3
40
@_AbhinavJ
Abhinav
5 months
had the opportunity to be one of the booth demos at the @socraticainfo symposium last night absolutely loved talking to everyone I met
Tweet media one
Tweet media two
Tweet media three
2
1
33
@_AbhinavJ
Abhinav
6 months
implemented SARSA and Q-learning from scratch. also its variants Expected SARSA and Double Q-learning. seeing the actual simulations of the optimal policies for all of these algorithms is amazing
Tweet media one
Tweet media two
0
1
9
@_AbhinavJ
Abhinav
6 months
been diving into reinforcement learning. too much theory lately, so I implemented policy iteration from scratch. many more optimizations can be made, but decent start
Tweet media one
0
1
14
@_AbhinavJ
Abhinav
7 months
I've been half assing effort for the past 27 days, every day I think today I'll finally get the ball rolling the way it was before, and here I'm at the same level of complacency weeks later. Gonna get back to when the winds of life didn't sway me as much.
0
0
4
@_AbhinavJ
Abhinav
11 months
I've spent the past three days, going in extreme detail about autoencoders and VAEs, deriving every mathematical expression along the way, and implementing each of the models in pytorch from scratch. did all this for a bigger project, should be posting about that soon
Tweet media one
Tweet media two
0
0
3
@_AbhinavJ
Abhinav
11 months
got my first co-op at Ford, rlly grateful to be done with the stress of job finding. worked at a cafe for the first time in kitchener, absolutely loved it. onto bigger projects now
Tweet media one
4
0
14
@_AbhinavJ
Abhinav
11 months
Completely forgot to post, 2 weeks ago, I attended my first hackathon, Hack the North, and made. Tenūto – an app that uses your heart rate and Spotify history to create personalized music therapy sessions, helping you relax and feel better over time.
Tweet card summary image
linkedin.com
Two weeks ago, I had the incredible experience of participating in my first hackathon, Hack the North—the largest in Canada, with over 1,000 hackers. Alongside my amazing teammates Allison Cretel,...
2
0
7
@_AbhinavJ
Abhinav
1 year
style-smart. an app which checks your schedule, the weather, takes the clothes in your wardrobe, and gives you an outfit for your upcoming day . cc: @_buildspace @FarzaTV
1
0
12
@_AbhinavJ
Abhinav
1 year
to clarify, I was training a model, and, I had the bounding box coordinates, so I added a 4th dim, corresponding to the binary mask for the box, then I augmented the training data, and to maintain the mask's position in geometric transforms I used the albumentations library.
0
0
2
@_AbhinavJ
Abhinav
1 year
i just spent 6 hours trying to experiment with a technique I thought would allow the model to better fit, only for it to completely fail. read so much on transformers, that I tried applying a binary mask as a way for some soft attention in a CNN. it's so bad, i'm crying inside.
1
0
4