Careers at Hedra: Teams, Pay and How to Get Hired

Hedra’s San Francisco listings show senior technical bands ranging from $175k to $325k (Zero G Talent board reported), reflecting competitive pay for San Francisco AI infrastructure roles.

Who gets hired and onto which teams

Hedra builds teams that combine broad marketing expertise with deep AI/ML talent, offering market‑aligned compensation and a hiring process that evaluates both technical and marketing fit.

What it pays

Hedra posts cash ranges that sit in the upper tier for San Francisco AI infrastructure roles. Its own listings show three senior technical bands that span the top of the local market. Together they span a wide band, with the upper end reaching $325k.

Equity details are not publicly disclosed. The LinkedIn note mentions “competitive compensation and equity” but does not list grant size, vesting schedule, or refresh policy. Levels.fyi’s October 2026 entry does not provide vesting schedule information.

The disclosed benefits include a 401(k), Silver PPO medical, vision, dental, and catered lunch plus snacks five days a week in the San Francisco office. Glassdoor’s benefits field reads “No benefits reported,” which stems from the low submission count. The in‑office requirement, five days a week, called out in the Distributed Systems posting, means the daily meal perk is tied to on‑site work.

When comparing offers, remember that Hedra prices senior engineering and research talent at the top of the San Francisco market, with posted bands that exceed most public self‑reports. The equity picture stays opaque until the offer stage, so ask directly about grant sizing and vesting. Treat the catered in‑office days as part of the total‑comp calculation.

How the hiring process works and what gets candidates through it

Hedra posts job descriptions that outline required experience and skills. For technical roles, these include experience with accelerated inference, pre‑training, post‑training, and building models that serve millions of users. For marketing and creative roles, they emphasize portfolio review, campaign metrics, and the ability to translate technical concepts for non‑technical audiences. The company’s own careers page states that it looks for people who care about accelerated inference and building models that reach millions of people.

Where the work happens

Hedra’s main office sits in San Francisco, with a second hub in New York, and the company employs fourteen people across both locations (builtin.com, 2024-12). This small footprint shapes how the space operates day to day.

Built In lists a typical on‑time of five days a week (builtin.com, 2024-12), whereas Hedra’s own builtinsf.com page says “Typical time on‑site: None”. This contrast shows differing expectations about attendance for the team.

Hedra builds visual models and the platform that serves them (image, video, audio, and vision‑language) at production scale (hedra.com, 2024-12). This work goes beyond experimental research; the company describes it as constructing the research and infrastructure behind visual intelligence.

Its flagship product, Character-3, calls itself the first multimodal model in production, covering generative images, video, and audio alongside other leading foundation models (builtinsf.com, 2024-12). This positions Character-3 at the forefront of released multimodal AI.

Hedra has launched the Hedra Creative Agent, a visual inference workspace that lets users run models without leaving the platform.

Job postings on Hedra’s own board show roles that optimize visual AI model inference across architectures, algorithms, runtimes, GPU kernels, and distributed systems. The work involves quantization, sparsity, caching, compilation, attention optimization, custom CUDA or Triton kernels, and multi‑GPU execution, done side‑by‑side with research scientists (builtin.com, 2024-12). This focus signals that Hedra expects its engineers to push model serving to the limits of hardware.

With a team of this size and a blend of marketing know‑how and solid AI/ML expertise, the San Francisco office works as a tightly integrated space where cross‑functional tasks happen out of necessity.

The Product Marketing Lead and Senior Video Producer roles listed on Hedra’s board show that marketing and creative production sit alongside the technical AI/ML core (Zerog Talent board data). This layout reflects the company’s stated goal of blending broad marketing know‑how with strong AI/ML capability.

The company’s priorities—ownership, fast ship cycles, and cross‑functional impact—shape how the San Francisco office operates.

With only this many people, the office cannot sustain bureaucratic layers. The ‘Balance and Joyful’ culture described in Hedra’s founding materials (hedragroup.com, 2022-07-28) tells us the physical environment supports work instead of blocking it. The ‘What Will You Create?’ message on the website points to a space built for creative output, not administrative overhead. Together, these cues indicate a workplace that favors flexibility and creation over rigid structure.

Although public sources give no exact square footage, layout details, or office amenities, the known data point to a flexible San Francisco workspace that can host the bi‑coastal team.

The back‑and‑forth over on‑site versus remote work mirrors a wider industry debate about how small AI firms arrange physical offices when their products are digital and distributed.

Joining Hedra’s San Francisco team means working in an environment where role boundaries, the line between technical and creative tasks, and the on‑site/remote split stay deliberately porous—an expression of the company’s priorities.

Team members may work on both technical and marketing tasks, reflecting the company’s cross‑functional focus.

The inference optimization work listed in hiring posts—quantization, sparsity, custom CUDA kernels—shows that the San Francisco office must run hardware able to handle distributed GPU execution and benchmarking workloads, even though Hedra’s own docs do not list exact cluster sizes.

This hardware runs beside the marketing and creative production teams that the main theme calls equally central to Hedra’s mission, forming a physical and organizational ecosystem where visual AI development and market‑facing product work share the same compressed, iterative space.

Who thrives here

Built In notes that Hedra’s culture emphasizes ownership, meritocracy, and hands‑on leadership.

Built In notes that Hedra operates with small teams, hard problems, and fast ship cycles, which shapes the workload and encourages taking initiative and seeing work through to users.

With a workforce that pairs broad marketing know‑how with solid AI/ML expertise, those who turn technical progress into market‑ready stories or who bring user insights back into model design are helpful and collaborative. Those who regularly sit in on both engineering and marketing discussions and who can speak both languages strengthen the team’s shared understanding.

Built In notes that Hedra offers unlimited paid time off and company‑sponsored outings, and that flexibility and shared team connection support work‑life balance.

Built In notes that the supportive culture includes free daily meals, snacks, and drinks that cut day‑to‑day friction and aid recovery after busy cycles.

Built In cautions that actual usage of perks can vary by team speed in a small, fast‑moving environment. Consequently, the team values those who communicate openly about their capacity and who adjust their workload to match team needs are valued. Those who speak up early about bandwidth concerns help the group maintain sustainable momentum.

If you recognize these traits in your own work style, you will find a fit where others notice your contributions and support your growth.

Working in frontier tech? Zero G Talent tracks the openings: see every open Hedra role, browse frontier tech jobs, the companies hiring, and the people building the field.