Neo AI
@withneo
Followers
10K
Following
960
Media
54
Statuses
278
First autonomous AI Engineer (34.2% on MLE Bench)
San Francisco, CA
Joined December 2022
Introducing NEO: The first Autonomous Machine Learning Engineer. It works like a full-stack ML engineer that never sleeps: handling data exploration, feature engineering, training, tuning, deployment, and monitoring, end to end. Powered by 11 specialized agents, NEO runs
28
57
365
There should be a new term to evaluate companies based on their impact on humanity, achievements vs just valuations… Even with similar valuations, SpaceX would be remembered 500 years from now for all the impossible made possible by them There should be a way to factor this in.
OpenAI is cool and everything but there's no way they are objectively worth more than SpaceX. They have one single hit product which is not and has never been profitable. If ChatGPT were to disappear tomorrow, it would suck but users could immediately pivot to several
1
0
3
We didn’t have $25k to buy a fancy booth at the AI conference. So we got creative and parked a truck outside the venue. Turns out the truck pulled more traffic than the booths inside, for a fraction of the cost. Moral of the story: why rent a table when you can own the street?
29
18
70
Congrats @satyanadella for hitting 30% and number 2 on the MLE bench. We’ve been holding the number 1 slot at 34.2% for 4 weeks now but competition keeps us sharp.
42
30
78
We aspire to be that interface through which you can learn and practice machine learning.
To all undergrads interested in learning about AI: be wary of taking “Intro to AI” as your first AI course. In many programs, the class you actually want first is “Intro to Machine Learning”. AI technology has exploded in the past 15 years thanks to deep neural networks. Yet at
2
2
9
10,000 agents in parallel building your models, maintaining them, improving them , is the new 10,000 hours
10,000 prompts is the new 10,000 hours
4
1
16
Decomposing word-word correlation matrix and computing non-negative matrices:
0
0
10
Improving the word error rate for whisper model on clinical speech dataset:
1
1
7
Doing sleep stage data exploration and feature analysis using a publicly sourced wearable dataset:
1
1
7
Watch NEO develop multiple ETA prediction models and perform comparative evaluation:
1
0
9
Get early access: https://t.co/SHJN4nqHna ✅ Neo is production-ready: ✅ Native integrations: Snowflake, Databricks, BigQuery, and more ✅ Deployable in your own VPC
1
2
15
A prime example of: You can just do things Microsoft scored 22.4%. NEO scored 34.2%. Imagine losing to 6 people, $500k in funding and some GPUs.
6
3
28
Everyone’s building ML like it’s still 2018. • Juggling data formats • Writing fragile glue code • Babysitting pipelines • Manually tuning for days • And hoping it all survives prod NEO doesn’t optimize around the chaos. It eliminates it. Not a wrapper. Not an assistant.
3
1
27
Do you know what 90% of ML work actually is? Not modeling. Not innovation. Just… cleaning data, debugging pipelines, re-running jobs, managing infra, and hunting for why the last output broke. We didn’t build NEO to write code. We built it to replace the grind. NEO is built
2
2
15
Do you know what happens when you give ML a Brain? You get NEO — the first Agentic ML Engineer. Not a co-pilot. Not a helper. NEO is powered by 11 specialized agents that plan, build, and ship entire ML pipelines — from raw data to deployment — without human stitching, guessing,
2
1
11