Partner @lightspeedvp, former early Ramp and Atrium, views are my own.
San Francisco
Joined June 2010
- Tweets379
- Following617
- Followers1.9K
- Likes1.6K
Lisa Han retweeted
There’s only so much AI can learn from the public internet. Some of the harder problems require data people have to create.
With Luel Labs, @LuelCompany is taking the human-data infrastructure it has built into multimodal research, starting with speech and underrepresented languages where existing datasets fall short.
We’re excited to see where @williamnamgyal, @Lenderking, and the team take Luel Labs.
Today we're launching Luel Labs, our end-to-end multimodal data lab.
Our research is powered by Luel's global network—leveraging 750,000+ contributors across 96+ countries to generate fully consented, quality-assured audio, video, and image data.
First up — Breakthrough TTS for underrepresented languages from the tips of the Himalayas and across the Sahara. We're delivering the first commercial-grade TTS for Tibetan and Dzongkha, and outperforming existing Masri and Darija models.
Alongside this effort, we’re driving foundational research to help the industry handle messy, real-world data—pioneering new approaches to 48kHz speaker separation, conversational overlap, STT modeling, data integrity, speech understanding and more.
RT @ravi_lsvp: @JeffDean has built the foundations that every era of modern computing has stood on, from web-scale infrastructure to scali…
This quoted post is unavailable.
Lisa Han retweeted
I deeply resent that AI has forced me to eliminate em dashes from my writing for fear of signaling slop
Lisa Han retweeted
Our first model, Inkling. Trained from scratch, weights are open, fine-tunable on Tinker today.
Today, we are introducing Inkling.
Inkling reasons efficiently across text, image, and audio modalities. We are making the full weights available.
thinkingmachines.ai/news/int…
Available today for fine-tuning on Tinker. Play with it in the Inkling Playground. 🧵
Lisa Han retweeted
Congratulations to @ravi_lsvp, @ravirajjain, and @buckymoore on their recognition in the Seed 100 List!
The Seed 100 List from @businessinsider highlights early-stage investors with a unique ability to scout the tech stars of tomorrow. Amid the AI boom, the competitiveness and speed of investors getting in before the “seed stage” as we know it have been reinforced.
This is the Seed 100’s sixth year, and it is an honor to have 3 Lightspeed team members acknowledged on the list.
Early-stage investing has been wired into our team’s DNA for over 26 years. And we are incredibly proud to have backed many teams from their Seed rounds and beyond.
As Ravi puts it: "The founders Lightspeed backs don't extrapolate from the present; they derive from first principles and arrive at futures others haven't thought to look for.”
Lisa Han retweeted
What started as two teenagers bonding over their love for digital gigs evolved into two founders turning their passion into a platform.
@luelcompany co-founders @williamnamgyal and @Lenderking built their digital marketplace to strengthen infrastructure for human data.
Luel invites AI labs and enterprises to submit dataset requests, match contributors to the task, and quality check each submission. Contributors are paid for their contributions, and AI companies gain valuable, real-world data that improves their software.
The embodiment of a symbiotic relationship.
That’s why we’re excited to announce that we co-led their recent $31.2M Seed round alongside @generalcatalyst.
There’s only so much data to be scraped online. The next AI models will need high-quality, human-led data based in reality, and Luel is among those leading the charge.
Lisa Han retweeted
I'm beyond excited to announce that I’m joining @Lightspeed as a Partner on the early-stage investing team and helping to build everything they're doing in new media.
I've spent the last six years investing in early-stage companies, and the last year building a media platform in parallel. What started as a side quest has reached 350,000+ followers and millions of viewers a month.
For me, investing and storytelling have always been two halves of the same job. This work has put me in rooms with incredible founders, researchers, and operators - many of whom have become close friends, collaborators, and early stage bets. There's so much happening at the frontier right now, and the media platform has given me a way to be in those conversations early (and to bring the people I admire most into a wider audience).
When Lightspeed reached out, we saw the same opportunity. Venture and media are converging, and the next generation of great firms will need to show up differently: reaching founders and builders before they've even started a company, and being the place where people first get excited about what's happening in tech.
With more than $50 billion in AUM, Lightspeed has been building alongside founders at the frontier of technology for decades. It's a firm I've admired for years since I started investing - for its intellectual rigor, its long-term conviction, and the way it approaches company building. I’m excited to join a team that brings real depth and substance to its work, and I hope to bring that same ethos to everything I put out.
Thank you to @ravi_lsvp, @bsomaia , and @Machiz for believing in this vision and trusting me to help build it. I’m so excited to see what we do together.
If you're an early-stage founder, building or thinking about starting something, or just curious about this ecosystem - please reach out. My DMs are open and I’d love to chat!
Lisa Han retweeted
The future of music is already here, and anyone can create it.
Today, @Suno announced a $250M Series C round to keep building an ecosystem where creators, listeners, and the broader music communities can make and share music together. Nearly 100 million people have created music on Suno in just two years, and many have found a new passion in the process.
Co-Founders @MikeyShulman, @fut_ure (Martin Camacho), @GeorgKucsko, and Keenan Freyberg are working towards building more advanced tools, richer experiences for casual creators, and new ways for people to connect through music.
Big congrats to the entire Suno team!
Cc: Lightspeed Partners @mignano and @lisabethhan
Lisa Han retweeted
Am I wrong in sensing a paradigm shift in AI?
Feels like we’re moving from a world obsessed with generalist LLM APIs to one where more and more companies are training, optimizing, and running their own models built on open source (especially smaller, specialized ones)
Some validating signs just in the past few weeks:
- @karpathy released nanochat to train models in just a few lines of code
- @thinkymachines launched a fine-tuning product
- rising popularity of @vllm_project, @sgl_project, @PrimeIntellect, Loras, trl,...
- 1M new repos on HF in the past 90 days (including the first open-source LLMs from @OpenAI)
And now, @nvidia just announced DGX Spark, powerful enough for everyone to fine-tune their own models at home.
Would you agree, or am I just seeing the future I want to exist? Also, why is this happening (just the advent of RL/post-training?)
Lisa Han retweeted
Dating advice: just do it
Selling advice: just do it
Writing advice: just do it
Coding advice: just do it
Fitness advice: just do it
Startup advice: just do it
Lisa Han retweeted
Today, we’re announcing our $1B Series C fundraise, and the opening of Factory 1.
It's time to rebuild America’s electric grid.⚡
Let’s go ⬇️
🤖 Made with AI
Lisa Han retweeted
Today we are announcing our $1B Series C, but money isn’t the story. The story is what we’ll do with it, and the scale of the challenge ahead.
Our mission isn’t constrained by capital, supply, or demand. It’s constrained by talent. To build America’s next great power company, we need the engineers, operators, and creatives ready to take on the challenge to join us.
Power is the most important product in the world. It runs through everything: our homes, hospitals, schools, factories, and future. And the grid that delivers it is breaking.
At Base, we’re building the future of power so human progress can accelerate. In the past two years, we’ve grown from an idea into one of the fastest growing energy companies in America.
Now we’re entering a new phase of growth. Our first manufacturing site is rising in the heart of Austin, inside the former Austin American-Statesman printing press. Here we’ll build our own hardware to meet surging demand and scale nationwide. This is a key step in putting a battery on every home in the country.
Few teams get to say their work powers human progress. We do. With a billion dollars of fuel, our next wave of growth will be defined not by capital, but by people who choose to step up, take ownership, and build the next great American power company.
The old system is breaking. Come build the new one.
Our Series C was led by Addition, with participation from both new and existing investors including Trust Ventures, Valor Equity Partners, Thrive Capital, Lightspeed, Andreessen Horowitz, Altimeter, StepStone, Elad Gil, 137 Ventures, Terrain, Waybury, Ribbit, CapitalG, Spark, BOND, Lowercarbon, Avenir, Glade Brook, Positive Sum, and 1789.
Lisa Han retweeted
Tinker is cool.
If you're a researcher/developer, tinker dramatically simplifies LLM post-training. You retain 90% of algorithmic creative control (usually related to data, loss function, the algorithm) while tinker handles the hard parts that you usually want to touch much less often (infra, forward/backward of the LLM itself, distributed training), meaning you can do these at well below <<10% of typical complexity involved. Compared to the more common and existing paradigm of "upload your data, we'll post-train your LLM", this is imo a more clever place to "slice up" the complexity of post-training, both delegating the heavy lifting, but also keeping majority of the data/algorithmic creative control.
I think the community still has to discover how and when finetuning makes sense compared to the (often strong) baseline of prompting a giant model. The early indications I've seen is that finetuning isn't so much about "stylizing" an LLM, instead, it's a lot more about narrowing the scope, and especially when you have a lot of training examples. An extreme example of scope narrowing being that of categorical classifiers, e.g.spam filters, content filters, etc. but it should be broader than that. Instead of building a giant few-shot prompts for a big LLM, it might work a lot better (and faster!) to finetune a smaller LLM specifically for your narrow task.
Increasingly, production applications of LLMs are larger pipelines where a bunch of LLMs collaborate in DAGs and flows. Some of these components might work well as prompts. But a lot of it will probably work a lot better as a finetune. Tinker makes the latter trivial and should allow for an easy experimentation of what works best at any stage.
Introducing Tinker: a flexible API for fine-tuning language models.
Write training loops in Python on your laptop; we'll run them on distributed GPUs.
Private beta starts today. We can't wait to see what researchers and developers build with cutting-edge open models!
thinkingmachines.ai/tinker
Lisa Han retweeted
Today we launched Tinker.
Tinker brings frontier tools to researchers, offering clean abstractions for writing experiments and training pipelines while handling distributed training complexity. It enables novel research, custom models, and solid baselines.
Excited to see what people build.
Introducing Tinker: a flexible API for fine-tuning language models.
Write training loops in Python on your laptop; we'll run them on distributed GPUs.
Private beta starts today. We can't wait to see what researchers and developers build with cutting-edge open models!
thinkingmachines.ai/tinker
Lisa Han retweeted
Excited to announce @lightspeedvp has partnered with @KodialamRo29640 @__jamie_b and team @getsphinx to bring true frontier intelligence to data teams.
Starting today, anyone can try Sphinx copilot for free. We are also hiring in-person in NYC. Please reach out if you want to be a part of changing the way we work with data.
sphinx.ai/blog/sphinx-launch…
Lisa Han retweeted
We’re proud to deepen our partnership with Lightspeed-backed @AnthropicAI in their $13B Series F financing round.
From the very beginning, Anthropic has had a clear strategy: build foundation models that are safer, steerable, and more interpretable. While others chased breadth, they doubled down on depth. That focus has positioned Anthropic as a key partner for enterprises exploring complex challenges in law, medicine, finance, coding, and more.
@claudeai Code is already showing that impact. Just two quarters after launch, it’s become a favorite tool for developers, with millions of weekly downloads and growing enterprise adoption.
Congratulations to Co-Founders @DarioAmodei, @DanielaAmodei, @nottombrown, @jackclarkSF, Jared Kaplan, @samsamoa, @ch402, and the entire Anthropic team on this milestone.
Read more: lsvp.com/stories/deepening-o…
Lisa Han retweeted
We @basepowerco are proud to be supporting families across Texas.
Lisa Han retweeted
New deep dive into what we're building @basepowerco by the team at @arenamag.
Excellent headline and excellent storytelling!
American Power is American Power ⚡️
The grid is aging. Energy demand is accelerating thanks to AI. The pressures on the system are immense, and power is becoming more important than ever.
Enter @BasePowerCo with a very *American* mission: more power, for less.
Lisa Han retweeted
New Geographies 🌎
Expansion outside of Texas is one of the key initiatives @basepowerco . We're hiring for an n of 1, first of a kind role that leads the operational effort of doing so.
If you're a self starter 0 to 1 type that likes to wear a ton of hats and build an operation from the ground up, apply at the link in the next post or DM me.
