filipe ✓

@filicroval · joined 05 May 2023

data eng | serendipity business inquiries: filipe@filicroval.com

153 508Followers

266Following

28 972Posts total

35.3KViews on collected posts

Ultimi post

filipe ✓@filicroval7 Sep

https://t.co/5DGxe3WJAF 1.1K views · 7 likes · 0 reposts · 1 replies Open on X →

filipe ✓@filicroval7 Sep

anyone using Hermes Bot and/or Grok Bot? what usecases do you have? i'm interested 1.1K views · 0 likes · 0 reposts · 0 replies Open on X →

filipe ✓@filicroval6 Sep

holy shit, openai confirmed they hit "automated research intern" level median researcher burning $600/day in tokens, 90th percentile at $7k/day. 3.1 agent-workdays for every human workday say what you want about openai but publishing internal RSI metrics (agent usage, task ht 2.5K views · 3 likes · 0 reposts · 0 replies Open on X →

filipe ✓@filicroval6 Sep

made my qwen3.8-27b dgx spark recipe 27% faster without touching the target model two changes, both to the speculative decoding side: 1. swapped the BF16 dflash2 drafter for maurienne-ai's calibrated NVFP4 one: the drafter runs every step 2. swept draft length. measured https:/ 5.8K views · 42 likes · 7 reposts · 2 replies Open on X →

filipe ✓@filicroval6 Sep

creating environments with GPT-6 Astra and Unreal Engine 5 is so fun it built this decaying greenhouse via computer use the workflow: → astra generates image references → unreal mcp to inspect and modify the actual scene → blender for custom geometry and damaged assets → https: 2.2K views · 0 likes · 1 reposts · 1 replies Open on X →

filipe ✓@filicroval28 Aug

Qwen3.8-27B now runs at 57.11 tok/s on a single NVIDIA DGX Spark here is a SGLang recipe for serving the stock Qwen3.8-27B NVFP4 checkpoint with DFlash2: → 12.34 tok/s without speculative decoding → 57.11 tok/s with DFlash2 → 4.63× median speedup across 18 performance rows ht 22.6K views · 145 likes · 15 reposts · 22 replies Open on X →

Rispetto ad account della stessa dimensione

6 post degli ultimi 90 giorni, accanto alla fascia di 100K–1M follower. copertura nella norma per la sua dimensione, reazione più debole della media.

Visualizzazioni mediane2 346questo account4 856mediana per 100K–1M

Copertura, %1.53%questo account1.80%mediana per 100K–1M

Interazione, %0.43%questo account1.32%mediana per 100K–1M

Metrica	Questo account	Mediana per 100K–1M	Rapporto
Visualizzazioni mediane per post	2 346	4 856	0.48×
Copertura (visualizzazioni ÷ follower)	1.53%	1.80%	0.85×
Tasso di interazione	0.43%	1.32%	0.33×

Altri account di questa fascia → Confronta con un altro account → Come sono costruiti questi parametri →

Growth & engagement

How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.

Views per post

22.6K28 Aug

2.2K6 Sep

5.8K

2.5K

1.1K7 Sep

1.1K

Last 6 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.

Engagement rate per post

0.83%28 Aug

0.09%6 Sep

0.90%

0.12%

0.00%7 Sep

0.74%

Reactions — likes, reposts, replies and quotes — divided by views. Median for 100K–1M accounts is 1.32%.

What the audience does

Likes65.9%197 in total

Reposts7.7%23 in total

Replies8.7%26 in total

Quotes2.3%7 in total

Bookmarks15.4%46 in total

Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.

The follower curve appears once this account has two daily snapshots — we take one a day, and this one is on its first.

Ultimi post

Rispetto ad account della stessa dimensione

Growth & engagement

Views per post

Engagement rate per post

What the audience does

Account simili