VP, Engineering @ Harvey. x-Meta/Uber/Twitter. | 🇦🇷·🇫🇷·🇺🇸 |📍🗽🏙️🍎🚕🍕
New York, NY
Joined November 2009
- Tweets7.7K
- Following851
- Followers8.7K
- Likes8.5K
Nico retweeted
LAB: US Case Law
We’re extending Harvey’s Legal Agent Bench to US case law, with 100 environments built from real federal dockets spanning 10+ areas of law.
Long-form writing matters for long-horizon agent research, especially in verticals with a significant degree of domain expertise like law.
Brief writing combines research, analysis, and drafting across hundreds of pages of source material. An agent has to preserve the evidence behind each claim through context compaction and use it correctly as it develops an argument. Errors during research or pre-writing phases compound and can carry over into the final draft.
Models are getting much better at producing complete first drafts, as evidenced in the video attached video, but there's still a lot of headroom to go!
Nico retweeted
Introducing Intake, where every ask to legal gets routed and resolved.
Requests come in through the tools your business uses, including email, Slack, Jira, and ServiceNow.
Intake gathers the right context, then routes and resolves requests using your legal team’s rules: where each request goes, what agents can resolve autonomously, and what needs a lawyer’s review.
Answers and follow-ups go back through the same channels.
Go from request to resolution faster with Intake.
Nico retweeted
Come help us scale @harvey’s model training team.
If you’re interested in bringing frontier agent research into the Harvey product and working with:
- @baseten to scale up RL to 80M+ token virtual datarooms
- @PrimeIntellect to create structured agent training environments from unstructured legal data
- @FireworksAI_HQ to navigate the quality <> cost Pareto frontier with inference-time routing and advisor models
- @LangChain & LangChain Labs to build efficient verifiers and close the observability <> training feedback loop
- @appliedcompute to post-train open weight models and high-volume agents for end-to-end legal tasks
- @EngramLab to create an entire synthetic law firm and firm knowledge memory systems for better / more efficient open-world search
- @trajectorylabs & @NVIDIAAI to shape the frontier of continual learning and sovereign AI for high-stakes domains
- @mercor & @SnorkelAI to build out Legal Agent Bench and other benchmarks across legal and other verticals
and other projects like this, then this is the role for you.
Apply here: harvey.ai/company/careers/d7…
We are hiring for @Harvey’s model training team.
This team will help Harvey expand from the application layer into the model layer and from legal into high end knowledge work more broadly.
We are hiring AI researchers of all seniority, particularly those with experience post-training frontier or open source models.
Our program is centered around large-scale model training, synthetic data generation, long horizon reinforcement learning, and rigorous evaluation in real world deployments. We are scaling-pilled and believe that nothing beats the combination of larger models and better training data.
We’ve been able to generate incredibly realistic legal environments and validated that this allows us to post-train open source models to achieve frontier performance with agents. We plan to scale up these data generation and training efforts significantly across legal to start, and eventually other verticals.
As a researcher, you will have access to thousands of GPUs and unique training data from our product and customer relationships.
Your research will inform Harvey’s product strategy and power AI used for some of the most economically and societally impactful work in the world.
Nico retweeted
Harvey is turning into a monster of a company.
Some highlights:
- Higher user engagement (53% DAU/MAU) than WhatsApp or Snapchat, on a B2B product for lawyers
- Added $100M ARR *this quarter*
- Now used by 75% of the AmLaw 100
- Added two major products to the suite in 12 weeks
- Unified product surface area within our cloud agent platform
- Continue to build out an insane executive bench (hired a CPO, CMO, CISO, and a CSO just in Q2)
- Some of our products have been growing 2-3x a quarter *for the last 6 quarters* - Vault uploads are up 280x+ since Jan 2025
Q2 recap for @harvey
- +$100M NNARR
- 53% DAU/MAU
Key hires (including Q1)
- Anique (CPO) - prev VP of Product at Rippling
- Rachel (CMO) - prev CMO at Notion
- Brooks (CISO) - prev CISO at Roblox
- Keith (CSO) - prev CPO at Google
Product
- Agent unification - cloud agents can use all Harvey product surfaces
- Command center (EA) - monitor adoption and ROI by use case
- Contract intelligence (EA) - agentic contracting platform for enterprises
Eng
- Migration to cloud agent infrastructure
- Integrating open source inference providers
- Scaling document processing (54TB / week)
AI
- Legal Agent Bench
- Open source post training
- Published multiple research directions with partners
We invested heavily in cloud agent infrastructure at the end of last year and in Q1. In Q2 we also unified many of our product surfaces (collapsed as @winstonweinberg says) by making them all tools accessible by our cloud agents.
Prior to this, there were a lot of capabilities in Harvey that were often only discovered by power users. As cloud agents get better and our product becomes more connected we are seeing users discover more of the product by learning from their agents (see plot of product surfaces per user).
👀
Q2 recap for @harvey
- +$100M NNARR
- 53% DAU/MAU
Key hires (including Q1)
- Anique (CPO) - prev VP of Product at Rippling
- Rachel (CMO) - prev CMO at Notion
- Brooks (CISO) - prev CISO at Roblox
- Keith (CSO) - prev CPO at Google
Product
- Agent unification - cloud agents can use all Harvey product surfaces
- Command center (EA) - monitor adoption and ROI by use case
- Contract intelligence (EA) - agentic contracting platform for enterprises
Eng
- Migration to cloud agent infrastructure
- Integrating open source inference providers
- Scaling document processing (54TB / week)
AI
- Legal Agent Bench
- Open source post training
- Published multiple research directions with partners
We invested heavily in cloud agent infrastructure at the end of last year and in Q1. In Q2 we also unified many of our product surfaces (collapsed as @winstonweinberg says) by making them all tools accessible by our cloud agents.
Prior to this, there were a lot of capabilities in Harvey that were often only discovered by power users. As cloud agents get better and our product becomes more connected we are seeing users discover more of the product by learning from their agents (see plot of product surfaces per user).
Nico retweeted
ok, here's how GLM 5.2 performs on a bench it definitely didn't see, and where GLM 5.1 scored 0.0%. Closer to Opus 4.8 than Sonnet 4.6
I hope their confidence seems more credible now
Vault went from 2.2M files uploaded per week in Jan to 15M in May, and just crossed 200M active files. The first 100M took 2 years. The second 2 months. Hugely proud of the team who pulled the work to make this work! harvey.ai/blog/faster-more-r…
Very happy to be working with you again, @cyanhex
when @ChristinaNWhite and I first led design together at @loom, I knew I’d found a forever friend + advocate / someone who knows how to build the conditions for great design
excited to be joining @harvey to learn from her + many others, and push myself during these wild times :)
Nico retweeted
Last week we held Harvey Hacks, our internal hackathon.
27 projects total across our 200-person eng team.
Wanted to highlight a few hackathon projects:
When you're building agentic AI for the most security-conscious enterprises on earth, you can't outsource your infra. Multi-model support, zero data retention, and cost control all point to owning the stack. I'm proud of the team making this possible!
Very exciting to see @harvey 's own @spencerpoff explaining how we test and push new @AnthropicAI models to their limit!
Expanding optionality for our worldwide customers. Particulièrement fier de cette release!
Now live in Harvey: Mistral.
Our platform brings together leading foundation models and routes work to the right one for each task.
Today we’re teaming up with @MistralAI to make their models available in Harvey, starting with EU customers.
Nico retweeted
gm
Native Project, Faulty Topography Theme, Frag 005
Expressing both Optics Logic and Phantom 3D gestures
Collected by overflowing.eth
Via @fellowshiptrust Daily Program Season 2
Nico retweeted
shadertoy.com/view/tcSyWK
The limit of the (1/n) double cusp groups is the apollonian group, but the limit of their limit sets is not the apollonian gasket. It adds an extra vertical translation symmetry to the tiling of the modular group \Gamma_2 in the upper half plane model.