"MiniMax" Wanted: From Diverse Backgrounds Across the Globe, United by One Mission | Oasis Vitality

Every day, we get a lot of messages across social media asking what it's like to work at MiniMax. The golden decade of mobile internet gave at least one generation a visceral sense of vitality and the feeling that they could change the world. But as that dividend approaches its ceiling, many are wondering: where's the next opportunity? What kind of people do I want to work with, and what interesting things can we build together?

Today, we asked some MiniMax teammates to share their stories. One of them, as an intern, mobilized the most compute resources in the company and led a team of more than a dozen people to train the first large model. Another walked away from a major tech firm, saying they could finally escape the five daily meetings and pointless internal competition to focus on writing code. Someone else found support they could never have imagined in academia, exploring new technologies freely and seeing real impact from their work. Another joined as an individual contributor and within six months became a technical leader in their own right, building voice models nearly indistinguishable from human speech. Still another received an invitation from across the Atlantic and decided to leave behind nearly ten years of life abroad to return to China and help build MiniMax's large model algorithms. What are their stories? Why did they join? What's the work culture like? Let's hear from them.

What's your MiniMax story?

• Pascal (Algorithm Team)

I studied in Europe and interned with a well-known European large model company's founder, where I clearly saw the trend: AI was moving toward training larger models with more data. After the internship, I returned to my lab to continue research, but was constrained by academic resources and couldn't work on the most cutting-edge, interesting problems. I had planned to pursue an industry-sponsored PhD to get more compute. Then, by chance, a headhunter and HR found me through the acknowledgments section of a paper and invited me to interview at MiniMax. After talking with the technical team, I found that everyone here was a smart, experienced AI scientist. No office politics — just pure focus on exploring frontier technology. So I decided to return to China and join.

I really enjoy MiniMax's pure startup atmosphere. Everyone works together to refine the technology and products, openly sharing technical details with each other to push algorithms forward. I hope more smart, hardworking people will join us, so we can all improve together through exchange, and make large model technology actually land in all kinds of scenarios. You'll never be bored here — you can expect all sorts of fun things to happen (like a hamster suddenly scurrying by during a meeting).

• Marco (Algorithm Team)

My research focus in school was natural language processing (NLP). I joined MiniMax in January 2022 as the company's first campus hire. At the time, I already had offers from several major Chinese tech companies through various "elite talent programs," and was ready to join one. But after a meal with Junjie (the CEO), I found that he had deep conviction about AGI and a clear vision for the technology and the company's direction. I felt that going to a big company would be too ordinary, without many opportunities or resources to build something from zero to one. Especially since GPT-3 had already come out — I believed the same thing would happen in China. So I came here.

When I was still an intern, I led the training of MiniMax's first-generation abab large model. It was practically an all-company effort, borrowing a dozen or so colleagues from various teams to build something from scratch. The results were stunning. From when I joined until now, our team has consistently used the most training resources in the company. For a campus hire to receive this much trust and room to operate — that's unimaginable anywhere else.

There are abundant resources and opportunities at MiniMax. If you can seize them, it's a great thing. I really hope people with an entrepreneurial spirit, who are dedicated, meticulous, and proactive, will join us!

• Huifeng (Data Analysis & Commercialization Team)

I experienced the high-growth period of a major tech company in 2019, when the whole place was full of energy. But by early 2023, the internet felt lifeless to me. As organizational mechanisms grew more complex and cross-department communication became harder, people's decisions increasingly focused on protecting their own turf rather than thinking about how to make the company better. I realized this wasn't the professional environment I wanted anymore.

Later, I chatted with several of MiniMax's founding team members and suddenly felt that rapid-climbing sensation again — everyone was talking about what would be cool to build, what was possible, even things that would be unique in the world. Then a friend I greatly admired (our product partner!) joined MiniMax, and I made up my mind to come too.

The people here are genuinely energetic. Everyone has lots of creative ideas, actively proposes solutions, and pulls others in to push things forward. They don't stop at empty talk.

At big companies, many people spend only 20% of their energy on actually doing things, and 80% on convincing stakeholders and pushing processes. A lot of people who join MiniMax have had experiences where they were ambitious but had nowhere to apply it. They come here hoping to do something valuable.

What these people have in common is that they look up at the stars while keeping their feet on the ground.

Looking up at the stars means having initiative, not passively being a cog. At a startup, there's a lot to do and many problems to solve. We hope you'll actively identify problems, take the lead on digesting them, and solve them.

At the same time, you need to be down-to-earth and focus on the work. We need every teammate to have strong execution and the will to push things through despite all obstacles. You may face uncertainty and need to handle some pressure, but at MiniMax, all kinds of possibilities and surprises can happen at any moment.

• Maple (Algorithm Team Intern)

I'm currently interning at MiniMax in the multimodal vision group. Besides helping the team with engineering implementation, I'm also doing some exploratory research with their support.

The biggest feeling from my internship is a shift in perspective. Compared to the unrestrained imagination in a lab, a company needs attention to practical details, seriously getting every seemingly small thing right. What I like most is the fast-iterating pace here — the work is intense but the team atmosphere is relaxed, with just the right amount of drive to help you improve quickly and get results.

Everyone here is reliable. Each person has their area of expertise and is very willing to communicate, which helps the whole group grow fast together. For example, before I came, my understanding of large-scale training was very limited. Within a week, I could see visible rapid growth. That's incredibly valuable.

• Huanshui (Algorithm Team)

I joined MiniMax in 2022 as employee #19. Before that, I spent three years doing speech synthesis at another company, but using traditional TTS technology — stilted, word-by-word, far from human-sounding. When large models emerged, speech had a new technical direction, and I was drawn to this trend and its possibilities, so I joined MiniMax.

We wanted speech synthesis to reach human-level quality. No one had walked this path yet. By late 2022, we had produced our first demo. The results weren't perfect, but we had validated the technical feasibility.

Both technically and organizationally, the company has grown very fast. Six months after joining, I built the speech synthesis team from zero to one, continuously iterating on large model capabilities. Today, the "Hailuo AI" you see that can clone a voice in 30 seconds is one of our team's achievements. This capability is now open to businesses and developers across industries, with AI quality, pauses, breathing, rhythm, and intonation all approaching human expression.

As for how we work and what it's like — taking the speech algorithm team as an example — our job is to find the certain solution amid all the uncertainty. There are many new directions, ideas, and papers now. We need the judgment to verify and implement feasible, high-performing solutions through experiments.

Engineering capability is also crucial. Our results don't stay in the lab — they need to genuinely improve model performance and be deployed in products. When we find the key to breaking through a performance bottleneck, we concentrate maximum force to solve it. That brings a lot of positive feedback and sense of accomplishment.

Also, MiniMax has a great internal atmosphere. Teams collaborate closely, pooling their wisdom to solve problems. There's no situation where everyone just handles their own patch, doesn't communicate, or even competes for work. Because what we do day-to-day is so new, we like to discuss and decide together when problems arise, rather than one person making a unilateral call and fighting alone.

Large models are a new赛道 — there's no perfectly matching major yet. So we care more about a teammate's inherent capabilities: whether they're grounded, pragmatic, and good at collaboration. When people ask what kind of people we're recruiting, project experience is definitely a plus, but it's not required. For example, we really like people with strong hands-on ability who are good at digging deep to solve problems.

What Kind of People Is MiniMax Looking For?

• A Dao (R&D Team)

I've been fortunate to catch two golden waves — PC and mobile internet. But from the second half of 2020, I clearly felt that internet growth had become exhausted, lacking that sense of doing meaningful things every day that defined the past decade. When MiniMax was founded at the end of 2021, most people didn't yet grasp the disruptive potential of artificial general intelligence, but MiniMax's founding team had very advanced technical judgment.** I saw the "startup atmosphere" and excitement I had missed, flew to Shanghai to meet the founding team in person, and joined a week later.

MiniMax has two core beliefs: talent-driven and technology-driven. On one hand, recruit excellent people, give them enough room, and maintain strong execution. On the other, keep the organization flat and agile, avoiding excessive hierarchy and information gaps when decisions need to be made.

Work here is relatively joyful because you're doing something socially valuable, learning a lot, and without the internal competition of big tech companies. Plus, you can quickly see progress and positive feedback — products and models iterate on a weekly cycle.

We hope future teammates believe in and endorse AGI. This赛道 is relatively new, and it's normal that many candidates haven't had deep exposure to it — we don't have hard requirements on that. We also don't care about age. But we hope you have excellent entrepreneurial spirit, professional ability, and learning ability, and that you stay attentive to knowledge and progress in the AI赛道.

• Gevin (Algorithm Team)

Let me briefly introduce myself: before joining MiniMax, I never thought I would develop my career back in China. All my previous study and work experience was in the United States. Since coming here, I've genuinely felt this is a cause worth dedicating ten or twenty years to. The people at MiniMax are true comrades-in-arms.

We believe that achieving AGI requires both top-tier scientific research and deep product integration, so that masses of users can use our models to meet real needs. Through interaction with users, the models keep getting better, thereby creating more value for users. We hope to "co-create" AGI with users through this cycle.

So we need more teammates to build better models together and realize the dream of AGI. You don't need prior large model R&D experience, but you need good learning ability and solid foundations in machine learning, computer science, and mathematics, so you can quickly pick up new technologies.

You need passion and imagination for AGI technology. That passion will drive you to constantly challenge unknown, revolutionary technical breakthroughs. You need a pragmatic character — the technical progress you pursue isn't just a theory or an experiment, but new technology that can truly change how millions of users live.

Everyone at MiniMax has their own story. What we share is our belief in AGI, our conviction in "Intelligence with Everyone" — using the best technology to serve every person. In less than two and a half years, we've grown incredibly fast — we've already launched self-developed general large models across different modalities, including a trillion-parameter MoE text model, a voice large model, and a vision large model. Based on our self-developed general models, we've also released two useful and fun native AI applications: Hailuo AI, which helps people improve learning and work efficiency, and STARFIELD, an immersive AI content community product. STARFIELD's monthly active users have already broken into the tens of millions!

Additionally, to let businesses and developers from all industries simply and efficiently build AI applications, we've launched the MiniMax Open Platform, providing secure, flexible, and reliable API services. It now serves more than 20,000 enterprises and individual developers.

But to achieve our AGI vision and co-create intelligence with users, MiniMax needs more teammates to join! Especially you — someone who can both look up at the stars and keep your feet on the ground.

At MiniMax, you will work on:

• Multimodal large models, large language models, and voice large model development
• Foundation model inference and training, algorithm engineering support, building the company's underlying development framework, and continuous optimization of technical infrastructure
• Building the best AI-native applications, so everyone can benefit from advances in AI technology
• Next-generation search, recommendation, and advertising integrated with large models
• FE, QA, architecture, web crawling, big data platform, and data warehouse

Compensation, Benefits, and More

Compensation: Highly competitive market rates

Benefits: Six insurances and two housing funds, 17–22 days paid leave, regular team building, annual salary review, housing subsidy, meal subsidy, and commute reimbursement

Flat internal communication and feedback mechanism: Weekly online "CD" meetings every Friday noon, where the founding team shares recent company updates, occasionally invites industry and technical experts to speak, and discusses operational improvements. Daily breakfast provided. Founders frequently treat everyone to milk tea, coffee, and snacks. (Yes, you will probably gain weight.)

*Portraits generated by STARFIELD, a MiniMax product