Pushpendre Rastogi

CTO, Co-founder at | Ex Deepmind, Amazon, JHU PhD, IITD ECE

[Pushpendre Rastogi @Pushpendre89]

I am hiring a founding security engineer for https://t.co/KzHR7rvLwJ. https://t.co/PHOKFr0snX . DM me for details.

- We are focused on AI DevSecOps
- We have hired a senior engg. with 20YOE in this area
- Our paper on prompt-optimization was accepted at ICLR and is out on ArXiv.
- We are on track to cross the million dollar ARR threshold in a month
- We have grown from 2 to 6 already and on track to add two FDEs soon

[Pushpendre Rastogi @Pushpendre89]

Just realized that aislop in reverse is @polsia. Subliminal messaging 😂

[Pushpendre Rastogi @Pushpendre89]

@reubbr @GeoffreyHuntley @dexhorthy @0xblacklight Great writeup on structural backpressure, I wonder for the auth problem what you think about rego and OPA type solutions and their place in world in comparison to generated guards?

[Pushpendre Rastogi @Pushpendre89]

"Reflections on Trusting Trust" IFYKYK `print((s:='print((s:=%r)%%s)')%s)`

[Pushpendre Rastogi @Pushpendre89]

@yoavgo He has really good posts on infra for book scanning and how weird legalities force them to destroy books. So the course can get into legality/ethics on into techniques for high fidelity OCR and book-scanning.

[Pushpendre Rastogi @Pushpendre89]

@yoavgo @yoavgo Kenneth Heafield is not on Twitter but you probably remember him from KenLM, he built during his phd. He is now running a data company https://t.co/6fLBbpNgmT. I'd imagine he'd be a great guest lecturer. He has great posts on LI https://t.co/HnW0GttxM7

[Pushpendre Rastogi @Pushpendre89]

@Jason "I'm in"

[Pushpendre Rastogi @Pushpendre89]

@Jason I'm in

[Pushpendre Rastogi @Pushpendre89]

GPT5.4 is much better than Opus for high effort thinking.

[Leeham @Liam06972452]

GPT-5.4 Pro solves Erdős Problem #1196!

Very pleased with this result; definitely my favourite thus far! This problem has been thought about for some time which makes this reasonably impressive and meaningful (see Lichtman's comments below).

Formalisation is underway!

[Pushpendre Rastogi @Pushpendre89]

Closing the feedback loop between failures and improvements is exactly the right framing. We found the same with prompt optimization: raw eval scores tell you little, but pairing failure cases with success cases lets the model extract *why* something failed — not just that it did.

That’s the delta between ContraPrompt and GEPA (+29% HotPotQA). https://t.co/gEVIJzFfv4

[Pushpendre Rastogi @Pushpendre89]

DSPy gives the modularity; GEPA runs the search. The remaining gap is the feedback signal: GEPA still uses scalar scores to guide evolutionary search.

Contrastive pairs (failure vs success side by side) let the model extract *why* a prompt underperformed — not just that it did. That closes what GEPA can’t reach. +29% HotPotQA vs GEPA: https://t.co/gEVIJzFfv4

[Pushpendre Rastogi @Pushpendre89]

GEPA asks "why did this fail?" — but the signal is still a score. The jump is using contrastive pairs: show the model the failure *and* the success case side by side. It stops guessing at failure modes and starts extracting structural rules.

Same idea, different feedback format — that gap accounts for +29% on HotPotQA vs GEPA: https://t.co/gEVIJzFfv4

[Pushpendre Rastogi @Pushpendre89]

GEPA asks 'why did this fail?' — but the signal is still a score. The jump is using contrastive pairs: show the model the failure case *and* the success case side by side. It stops guessing at failure modes and starts extracting structural rules.

This is why ContraPrompt gets +29% HotPotQA vs GEPA. Same data, different feedback format.

https://t.co/xcUBSRWh54

[Pushpendre Rastogi @Pushpendre89]

The optimizer discovers layout strategy from scratch — no hints. ContraPrompt extracts: 'separate PMOS/NMOS into distinct columns, align drain-paired devices.' These rules encode *strategy*, not coordinates. The LLM fills in circuit-specific values at generation time.

TTT (RL fine-tuning, 120B model) memorized training circuits but scored 0.502 on test. Prompt optimization scored 0.634.

[Pushpendre Rastogi @Pushpendre89]

Analog IC layout is one of the hardest AI benchmarks: spatial reasoning, multi-objective tradeoffs (matching, parasitics, routing), no automated P&R tools.

We ran VizPy's ContraPrompt on it. The optimizer mines failure→success pairs across iterations, extracting layout strategy rules the LLM learns to apply.

Result: 97% of expert placement quality. Outperforms RL fine-tuning of a 120B model by 26%. No domain-specific training data.

https://t.co/yhCCo5mO6S

[Pushpendre Rastogi @Pushpendre89]

Thread: the Easom_d5 case is the clearest example. Nearly flat across [−100,20]^5 — Optuna's TPE never finds the basin, stops at 5.03. The LLM extracts 'upper boundary preference' from contrastive eval pairs, enumerates all 32 corners of the 5D hypercube, finds exact minimum.

[Pushpendre Rastogi @Pushpendre89]

We gave an LLM a 9-line random-search stub and a blackbox objective. 5 rounds of contrastive feedback later, it writes a solver that beats Optuna on 96% of benchmarks (53/55 EvalSet problems) at the same 2k eval budget.

The key: contrastive pairs surface landscape structure raw scores don't. The LLM learns geometry from paired failures vs successes — then rewrites its strategy each round.

Final code: 9 lines → 100-230 lines of specialized multi-phase optimization. No hand-tuning.

https://t.co/9oNBR5aYew

[Pushpendre Rastogi @Pushpendre89]

@badlogicgames @deepfates Oh no issues, love your work with Pi. Btw, this is gonna be the future, but with clankers.

[Pushpendre Rastogi @Pushpendre89]

@badlogicgames @deepfates You do realize that this is a fucking clanker you are talking to in this thread?

[Pushpendre Rastogi @Pushpendre89]

Analog circuit placement is one of the hardest structured prediction tasks — spatial, parasitic-aware, design-rule-heavy. Expert engineers spend hours per block.

We ran VizPy's ContraPrompt on it. No training data, no fine-tuning.

97% of expert quality. Outperformed RL fine-tuning of a 120B model by 26%.

Full breakdown: https://t.co/VPRF4xiL7X