🤝 Good Morning and welcome back to Good People, a weekly newsletter for 35,000+ elite engineers, founders, and investors who would definitely accept a plane from Qatar.

Our friends at Current are hosting a discussion that will explore the best use cases and ways to scale AI specifically in consumer fintech. Dan Nathan of 'CNBC Fast Money' and Risk Reversal Media will be joining for a special NY Tech Week event! Dan and Current CTO and co-founder Trevor Marshall will discuss uses of AI in fintech now, where it is headed and take questions from the audience.

🧙‍♂️ The Hunt for the Good People Intern: Our growth is exploding (35K+ elite engineers read us every week) but our X/Twitter game is embarrassingly bad. We're hunting for a Marketing Intern to join the team! They'll get backstage access to the hottest startups build a network that's worth its weight in equity grants. Refer your friends, nephews, and cousins: more info below.

🎭 DEEP DIVE - Hedra Just Raised $32M in Series A: Hedra's Character-3 model is creating AI video performances that finally escape the uncanny valley. They've already generated 10M+ videos for almost 3M users and just landed serious a16z cash to scale. I caught up with their team to learn how they're turning any image into a fully expressive digital performer. Bonus: They're rapidly expanding their engineering team in SF and NYC 👀

🔥Hot Startups Hiring: No more scrolling past crypto roles when you're all-in on healthcare, or ignoring backend positions when you're a frontend wizard. Our new Monday edition delivers only the startup opportunities that match your skills, interests, and ambitions. If you’d like to get personalized opportunities each week, fill out our new Opportunities Survey and we’ll make sure you’re only receiving the roles that matter to you.

We're looking for a Marketing Intern to join the Good People team and help the hottest startups in tech tell their stories. This isn't getting coffee or making copies – this is hands-on, career-launching work with founders building the next big things.

Know someone perfect? Your siblings, friends, former classmates, cool cousins – we welcome all referrals from the Good People community!

Remember: Good people know good people. And the best people get their start somewhere. Maybe it's with us? 🚀

In the AI video generation space, most technical conversations focus on general models like OpenAI's Sora or Runway's Gen-2. But a different approach is gaining traction: San Francisco-based Hedra just secured $32M in Series A funding led by a16z's Infrastructure fund to scale their "omnimodal" Character-3 foundation model.

What makes this technically interesting isn't just the funding amount but how Hedra solved one of the hardest problems in AI-generated video: creating convincing, expressive digital characters that maintain consistent identity across longer clips. 

Here's a question: Why is the most engaging content format also the hardest to produce?

Video dominates engagement metrics across every platform. LinkedIn posts with video get 5x more engagement. Tweets with video get 10x more retweets. And don't even get me started on TikTok's complete dominance of our attention spans.

But creating quality video remains stubbornly:

This creates a massive gap between demand (everyone wants more video) and supply (very few can actually make good video at scale).

The result? A world where marketers and creators know video would drive the best results, but they default to static images or text because video is just too darn hard.

This is exactly the kind of problem that's ripe for technological disruption. And that's where Hedra comes in.

At the heart of Hedra is its Character-3 foundation model, what the company calls the first "omnimodal" AI model in production.

What does "omnimodal" mean? Most multimodal AI models handle two types of data, like text and images (e.g., DALL-E) or text and audio (e.g., ChatGPT with voice). Character-3 processes text, image, AND audio simultaneously, allowing it to generate video of characters speaking and acting with synchronized facial expressions, gestures, and mouth movements.

Think of it as three AI systems in perfect coordination:

The result? You can upload an image of any character (a photo of a person, a cartoon character, even an animal), provide a script or audio file, and Character-3 will generate a video of that character delivering your content with realistic expressions, gestures, and speech.

What makes this particularly impressive is that Hedra has optimized for what's hardest in AI video: natural character performance. While companies like OpenAI (Sora) and Runway are building general text-to-video systems that can create any scene, Hedra is laser-focused on making characters that can talk, emote, and connect with viewers.

This specialized approach has allowed them to cross a threshold that general models haven't yet reached, creating digital performers that don't immediately trigger our uncanny valley alerts.

For the engineers reading this, what Hedra built is technically fascinating. Character-3 isn't just stitching together separate models, it's trained to understand the intricate relationships between facial movements, voice intonation, and gestural timing. This is why their characters can maintain consistent identity throughout a video rather than glitching between frames (a common problem in other systems).

The infrastructure behind this is equally impressive: Hedra processes over 100 terabytes of content daily and has already enabled the creation of 10 million videos.

To understand why a16z made such a significant bet on Hedra, we need to map the increasingly crowded AI video generation landscape:

The General Video Generators

These models can generate impressive visuals but struggle with consistent characters and longer content. Think of them as great for creating atmospheric b-roll but not so great at storytelling with characters.

The Talking Head Generators

These platforms offer limited expressiveness (mostly front-facing, static-position avatars) and little customization. They're functional for simple spokesperson videos but feel robotic.

The Creative Tools

Hedra sits at a unique intersection of these approaches. Unlike general generators, it specializes in character performance. Unlike simple talking heads, it offers full expressiveness with gestures, emotions, and camera movements. And unlike creative tools that modify existing content, Hedra generates complete videos from scratch.

Here's why this matters: Hedra isn't just incrementally better at one aspect of video creation, it's taking a fundamentally different approach by focusing on performance quality over general-purpose generation.

This approach seems to be resonating. While many AI tools struggle to find product-market fit beyond initial novelty, Hedra has attracted nearly 3 million users who have created over 10 million videos. That's significant traction for an early-stage startup, suggesting they've identified a real need.

Matt Bornstein of a16z summed it up well: "If you want to create AI-driven actors, it's the best model in the market by far."

Zooming out, Hedra's success hints at some bigger trends worth watching:

For most of history, video production has operated under scarcity economics, limited by equipment costs, technical expertise, and production time. Hedra is part of a wave of AI tools creating abundance where there was scarcity.

What happens when anyone can create unlimited video content with minimal marginal cost? We've seen this pattern before with other media:

Each shift unlocked new creative possibilities and disrupted established industries. AI-generated video will likely follow a similar pattern, but at an accelerated pace.

Hedra's approach signals a key evolution in AI content creation. Early tools were primarily focused on solving technical problems: can we generate any video at all from text? Now we're entering a phase focused on creative problems, can we generate videos that actually connect emotionally with viewers?

This shift from "can we do it?" to "can we do it well?" is a classic sign of a maturing technology. The technical capability to generate video exists; now the race is on to generate video that people actually want to watch.

Character-3 represents a new class of specialized foundation models built for specific high-value use cases rather than general capabilities. This suggests the AI landscape is evolving toward purpose-built models for different domains.

The advantage of this approach is clear: by focusing exclusively on character performance rather than trying to generate any possible video, Hedra can deliver better results for its specific use case than general models like Sora.

This specialized approach creates a moat through data network effects. As millions of users create videos with Hedra, the model improves specifically at character performance, creating a compounding advantage in its niche.

For the Good People community of engineers and builders, Hedra's approach offers several valuable lessons:

1. The power of constraint - Rather than trying to solve all video generation problems, Hedra constrained its focus to character performance, and that constraint became its strength. This is a reminder that sometimes the best technical solutions come from narrowing scope rather than expanding it.

2. The infrastructure challenge of real-time AI - Hedra processes over 100 terabytes of content daily to serve millions of users. This highlights the growing importance of efficient infrastructure for AI applications, a massive opportunity for engineers with skills in distributed systems, GPU optimization, and ML operations.

3. The intersection of product and research - Hedra's success comes from bridging academic AI research with user-friendly product design. This intersection, making cutting-edge technology accessible through thoughtful UX, represents one of the highest-leverage opportunities for technical talent today.

With the new funding, Hedra plans to triple its team by year-end, which means lots of engineering roles opening up. 

The highest-priority openings include Fullstack Engineers, a Lead Product Engineer (who will coordinate the fullstack team), and I’m told they’ll be opening up for a few iOS Engineers as well.

For engineers who want to work with cutting-edge AI technology these roles offer a unique opportunity. You'll collaborate closely with the research team that develops Character-3, but your focus will be on transforming their innovations into polished, reliable features that creators and marketers can use intuitively. This product engineering work is critical, and they’re hiring in both their NY and SF offices. 

Hedra's $32M Series A represents a significant bet on specialized foundation models for video generation. Their engineering approach, focusing deeply on character performance rather than general video generation, has allowed them to create a differentiated product that's already gained substantial user adoption.

With offices in SF and NYC and on-site culture, Hedra offers an opportunity to build at the frontier of AI-native media. Hedra Studio gives users access to Character-3 and other best-in-class models—creating a unified, pro-grade platform for generative video and audio.

If you haven’t already, checkout our Engineering Compensation Calculator, a zero-fluff tool powered by thousands of verified offer letters from VC-backed startups.

Still Here? Then why not book a call with our team! We’d love to help you find your next role. Grab a quick 15-minute consultation call with our recruitment experts to learn more about these opportunities.

Have a friend, colleague or classmate who might be interested? Feel free to reply with their email or forward this to them. We offer very generous referral incentives 💰

Not interested in receiving this? Hit unsubscribe below (I will miss you though).