Nenad Mancevic @manca1

Principal Engineer @HBOMax, architecting massive-scale Video AI & empowering engineers. Roots in hard real-time video streaming & Computer Vision. Views my own. mancevic.com San Francisco, CA Joined June 2007

Tweets

1K
Followers

393
Following

213
Likes

479

Nenad Mancevic @manca1

2 days ago

Teenagers have likely replaced reading with social media. I’m hearing more and more middle schoolers are constantly on YouTube, TikTok and Roblox. These trends are very concerning.

Bird on Fire @nobodyknows2322

5 days ago

Teenagers have likely replaced reading with social media. I’m hearing more and more middle schoolers are constantly on YouTube, TikTok and Roblox. These trends are very concerning.

207 843 5K 4.4M 897

Download Image

0 0 2 151 0

Nenad Mancevic @manca1

4 days ago

High-quality data created by domain experts is the new oil. There are two major benefits to having ultra-high-quality data in the AI-first world: 1. LLM post-training - aligning model outputs with human-level expertise. 2. Evaluation and benchmarking - measuring LLM…

0 1 2 199 0

Nenad Mancevic @manca1

3 days ago

Looks familiar? AI Bubble vs the Dotcom Bubble Maybe history does repeat itself

0 0 3 135 0

Download Image

Nenad Mancevic @manca1

5 days ago

With the recent buzz about hundreds of billions of dollars being poured into AI infrastructure projects, it’s hard not to draw a parallel to the early 20th century, when the world was being wired for electricity. We’re living through a truly unique moment in history.

0 0 2 140 0

Nenad Mancevic @manca1

7 days ago

Of course, the infinite loop goes like this: 1. NVIDIA invests $100B in OpenAI 2. OpenAI buys more NVIDIA GPUs 3. NVIDIA market cap increases by $100B 4. Repeat until singularity is reached.

0 1 2 270 1

Nenad Mancevic @manca1

7 days ago

How about $100B for OpenAI? NVIDIA is on a roll!

Nenad Mancevic @manca1

2 weeks ago

How about $100B for OpenAI? NVIDIA is on a roll!

0 0 2 517 0

0 0 2 227 0

Nenad Mancevic @manca1

a week ago

Evals and RL environments are the new hot topic in Silicon Valley, but the core ideas aren’t really new. In supervised ML, we’ve always needed a way to measure model quality. That means defining metrics (precision, recall, F1, BLEU, etc.), curating a dataset with high-quality…

0 0 2 169 1

Nenad Mancevic @manca1

a week ago

Math lesson for a 6yo today: heap trees and binary search trees.

0 0 3 207 0

Nenad Mancevic @manca1

2 weeks ago

Been working on an enterprise-grade eval framework for different data types lately. Metrics like BLEU, ROUGE, precision/recall only go so far. LLM judges are promising, but their scores can be hallucinated. Feels like the real SOTA is still HITL, but it’s not scalable. WDYT?

0 0 3 395 1

Nenad Mancevic @manca1

2 weeks ago

Cursor and the likes can seriously erode the swe craft and problem solving skills. If you blindly throw them a problem and just take the solution they spit out for granted, your problem solving muscle will atrophy and you’ll inevitably become dumber.

Jack Sharkey @jacksharkey11

2 weeks ago

475 192 4K 904K 985

0 1 6 1K 2

Nenad Mancevic @manca1

2 weeks ago

Hot take: most “AI-first” initiatives will flop. But we’re still in the early innings of enterprise AI. The few who persist will reinvent themselves and leave everyone else behind.

Nenad Mancevic @manca1

a month ago

Hot take: most “AI-first” initiatives will flop. But we’re still in the early innings of enterprise AI. The few who persist will reinvent themselves and leave everyone else behind.