Siddish

now: premise bio; ex- first ai engineer @metaformsai (evals, agents)

[Siddish @siddish_]

Sooo we recently announced our $9M Series A 🤗

@metaformsai 🐘 is growing bigger

[Peak XV Partners @peakxvpartners]

Traditional market research systems are outdated, tedious, and cannot scale while the demand from businesses is increasing. @metaformsai is changing this by helping research agencies become AI-first.
Excited to join the mission w @arjunsme & @ofcAkshat🤝.
https://t.co/Eq0rp52maj

[Arnav Bansal ⠕ @itsarnavb]

If you're NOT in Bangalore and working on the public good on any of our favorite problem areas, come to Bangerlore

We at @_lagrangepoint will help with flights and accommodation, and help you meet fellow practitioners. Just DM me!

- urban quality of life / civic tech
- food toxicity
- deep tech talent

If you know someone that fits, send them my way, and I will gift you some really nice socks

[Dhruv Trehan @dhruvtrehan9]

⏰ 3 days left until the deadline for the Conference for AI Scientists (CAISc) 2026.

With awards incl. model and compute credits over $20,000, this is a great chance to get feedback on ongoing research.

Our ask - use an LLM or AI system in some part of your research workflow.

[Snake Oil Salesman @ghuubear]

built this in the last 4 weeks or so. it's a full fledged market research delivery ops simulator with operations, finance, demand-gen, and HR modeled in. wanted a good way to teach operational concepts + simulate what agents do to an operational chain
https://t.co/98lvlhjngq

[Siddish @siddish_]

@ghuubear woahhhh
this brings back stoa memories too

[Siddish @siddish_]

@anindyadeeps @YourStoryCo @try_litefold @cory_jay44 Less goo brother!

[Anindyadeep @anindyadeeps]

LiteFold just got featured in @YourStoryCo

It’s been 8 months since Cory and I officially started @try_litefold.

We began by building better interfaces and workflows for computational biology, followed by Rosalind, our AI Co-Scientist for therapeutic design.

Now we’re gradually moving toward developing our own in-house scientific and AI research focused on next-generation therapeutic models and research environments and partnerships.

Still very early. Really cant wait to share whats coming.

Story link in the comments

[Adithya S K @adithya_s_k]

Excited to release the Ultimate guide to RL environments!

Definitions of RL environments differ wildly in the LLM era, so we spent the last month building several RL environments across 6 different frameworks, domains and complexities to map out which are easiest to build with and which can be scaled to 1000s.

[Siddish @siddish_]

@akashchanran You got taste brother

[Siddish @siddish_]

@akashchanran Wow, setting standards with this demo audio

[RL Commons @RLCommons]

Introducing RL Commons.

An open research initiative for the reinforcement learning era.

Shared compute. Open benchmarks. Collaborative work.

Founding phase Project Aster is now open.

[Ozzie @ozziekirkby]

Frontier LLMs can do a lot—but can they write good flashcards?

Turns out: not yet! @andy_matuschak and I created an eval for flashcard generation and found surprisingly poor results.

Worse, newer models aren’t helping: GPT 5.4 performs worse than 5.2, Opus 4.7 worse than 4.6.

[Siddish @siddish_]

@soham_btw what!? crazy

[harry sanders @harrysanders22]

this is our thesis. it's about squid, MRI machines, and why we think math is what's holding science back.

https://t.co/4Xv3VB69Qp

[Siddish @siddish_]

@lifeofadvait @AshwanthramKL

[Siddish @siddish_]

@asvs_kartheek ahh, this is a thin in-house harness

nothing fancy, just wanted some tools to be run on the users browser/extension and some reliability via model fallbacks, end user in the loop etc

[Siddish @siddish_]

./skills files are just directed graphs. took our skills hitting 300k+ tokens and agents consistently missing discovery and latency issues. so

1. depth should be inverse of frequency. stuff needed every run → 1 hop from root. niche edge case → 3 hops deep is fine.

[Siddish @siddish_]

@asvs_kartheek like a git undo: revert(steps: int, message: str)

as this tool call is not specific to undoing skill file reads, whenever agent attempts something and fails like search_replace on artifacts it can decide to revert

this mattered when models were like just 150k usable contexts.

[Siddish @siddish_]

some of them sound like our harness problems, have to run some rlm experiments

[Siddish @siddish_]

3. but every hop is a file read and every file read costs context window. wrong file = wasted tokens. graph structure is literally inference cost optimization.

we added a tool call that undoes a file read and redacts it from context before compaction hits