Interested in all things RL, Cricket, Badminton and real-time systems.
Joined February 2017
- Tweets 486
- Following 251
- Followers 83
- Likes 3,124
21 Photos and videos
Jun 11
🤣🤣
11
Jun 9
ozzy my man, hope you're chilling out there
thanks again for all the incredible music
open.spotify.com/track/755bv…
14
May 16
Some data during the era of slop (guilty as charged, having produced some myself, realizing starkingly now after having spent several hours fixing some of my recent code).
"The slop culture time bomb"
open.substack.com/pub/riskya…
53
May 13
the history gladly repeats.
4
161
May 9
Wrote this slowly but finally.
The Chaos at Our Doorstep - Why India Must Centralize Before It Can Decentralize.
open.substack.com/pub/riskya…
41
Mar 29
my goodness, I had to verify if this is real or not and this sadly very much is.
Mar 28
🚨 UNREAL
84 yr-old went to a Vancouver hospital with back pain.
Before anyone even told her what was wrong, a doctor offered her MAID.
She had a fractured sacrum.
A broken bone.
She said no.
Six weeks later she walked her daughter down the aisle.
Then she climbed a volcano in Guatemala.
Her words: "I did not want to die. I just wanted to find out why I was in pain."
In Canada, they offer to help you die before they offer to help you heal.
nationalpost.com/news/canada…
73
Mar 27
We used our optimizer to generate solutions to black-box functions and as it turns out -- it does pretty well starting with a very basic random search stub.
Mar 26
We gave an LLM a 9-line random-search stub.
No solver logic. No domain knowledge. Just a seed and a scoring function.
Five rounds of contrastive feedback later, it wrote its own optimizer that matched or beat Optuna on 53 out of 55 benchmark problems. On 39 of those, it found the exact global minimum.
vizops.ai/blog/contraprompt-…
73
Mar 19
One of my favorite works. We are working to ensure our optimization engine can *optimize everything* -- I believe we need to hit the ceiling of non-parametric optimization before RL and this is what has motivated our recent work to extend our optimization engine to several new applications.
Mar 19
We taught an LLM to place analog circuits with no training data, no fine-tuning.
97% of expert quality. One pass.
Full write-up: vizops.ai/blog/prompt-optimi…
Here's what happened 🧵
1
1
83
Rishav retweeted
Mar 10
Vizpy is live on producthunt now.
producthunt.com/products/viz…
Vizpy is a state-of-the-art prompt optimization service which learns from failures and updates your prompts with rules that it has learned. We have compared it extensively against baselines such as GEPA on benchmarks like BBH, HotPotQA, GPQA Diamond, and GDPR-Bench and VizPy wins on all of them. We'll have more benchmarks on cyber-security and chip-design coming out soon.
@producthunt
1
1
138
Mar 10
We are live on product hunt now!
Mar 10
We’re live on Product Hunt 🚀
VizPy automatically optimizes LLM prompts by learning from failures.
One API call, dramatically better results.
Would love your support and feedback: producthunt.com/products/viz…
41
Rishav retweeted
Mar 10
We’re live on Product Hunt 🚀
VizPy automatically optimizes LLM prompts by learning from failures.
One API call, dramatically better results.
Would love your support and feedback: producthunt.com/products/viz…
4
1
5
97
Rishav retweeted
Mar 8
Most people confuse prompt engineering with prompt optimization.
They're not the same thing. Not even close.
Here's why it matters more than you think 🧵
1
1
2
42
Rishav retweeted
Mar 4
🧵 1/ We are excited to announce VizPy — the latest product from Vizops.
VizPy is a drop-in replacement for DSPy. It's a prompt optimizer that finds where your AI breaks and fixes it automatically.
GEPA set the bar. We built VizPy to beat it.
👉 vizpy.vizops.ai
ALT VizPy Landing Page Image
1
1
1
120
Mar 2
Our work is out for use in production.
Its beats the current state-of-the-art, verified by OpenClaw 🙂
A few weeks ago I had hinted at a new prompt optimizer service that beats Gepa!!! We are live now!! The setup is so simple that openclaw can install it and test it for you.
Prompt: "Go to vizpy.vizops.ai, describe their service, set up an experiment comparing GEPA from dspy vs their optimizer. Does it actually work?"
51
Rishav retweeted
A few weeks ago I had hinted at a new prompt optimizer service that beats Gepa!!! We are live now!! The setup is so simple that openclaw can install it and test it for you.
Prompt: "Go to vizpy.vizops.ai, describe their service, set up an experiment comparing GEPA from dspy vs their optimizer. Does it actually work?"
1
3
448
Rishav retweeted
Feb 23
Distillation does have significant impact!
Feb 23
We’ve identified industrial-scale distillation attacks on our models by DeepSeek, Moonshot AI, and MiniMax.
These labs created over 24,000 fraudulent accounts and generated over 16 million exchanges with Claude, extracting its capabilities to train and improve their own models.
34
136
2,608
163,423
Feb 19
hang in there for some cool work :)
coming soon.
Feb 19
Great post from @clattner_llvm on Anthropic's C Compiler. Chris's work on LLVM obviously needs no introduction. it's mentioned in all three of the compiler books that I own :)
Also happy to see him citing some of the work that we open-sourced at @vizopsdotai . Btw, we'll soon be releasing a new prompt-optimizer that we built (thanks to @rishav_real). It beats Gepa on a bunch of benchmarks. Watch this space.
1
88