Benjamin Reavill

Multimodal LLM Researcher$300,000 - $400,000 Remote, Palo AltoFull-time / PermanentDeepRec has partnered with a high-growth generative AI company (Series B, $130M raised). They're building multimodal, multi-agent systems that combine language, vision, audio, and video. If you've been looking for a role where your research reaches production and shapes how millions interact with creative AI, this is worth a closer look.You'll help define the next generation of multimodal AI systems. Your work will span research, experimentation, and deployment, with a focus on real-time performance, multimodal reasoning, and agent-based workflows. You'll have the freedom to explore ambitious ideas while working alongside engineers who can bring them into production.What You'll Do- Lead research across LLMs, VLMs, and Audio Language Models- Design novel multimodal model architectures and training approaches- Improve real-time inference across text, image, audio, and video- Train and fine-tune autoregressive and diffusion models- Build and curate high-quality multimodal datasets- Collaborate with engineering teams to deploy research outcomes- Publish findings at leading AI conferences and journalsWhat You'll BringEssential- Strong research track record in multimodal AI or foundation models- First-author publications at recognised ML, vision, or audio conferences- Deep expertise in LLMs, VLMs, Audio LMs, or related fields- Strong Python and deep learning experience using modern frameworksDesirable- Experience with diffusion models or world models- Background in real-time AI systems and model serving- Experience building large-scale multimodal datasetsWe encourage you to apply even if you don't meet every requirement. The right mindset matters as much as the right CV.What's In It For You- USD 300,000–400,000 salary- Fully remote working arrangement- Ownership of research that shapes production systems- Opportunity to publish and contribute to the field- Direct collaboration with product and engineering leadershipThis role offers the chance to work on multimodal AI problems that sit at the intersection of research and real-world deployment. If you're excited by advancing the field while seeing your work reach users, we'd love to hear from you.

Senior Agentic AI Engineer$300,000 - $400,000Remote / Hybrid, Palo Alto Full time / PermanentA well-known, frontier GenAI company (Series B, $130M raised) is undergoing a major product pivot. They're moving from single-modal generative experiences toward a consumer-facing, multimodal, multi-agent ecosystem designed to feel genuinely autonomous, useful, and alive.They’re building the core infrastructure that will define how millions of users interact with AI agents daily. From planning and execution to memory, creativity, and proactive behaviour. This role sits at the heart of this shift: designing and shipping the systems that make intelligent agents function for 1M users.What You’ll DoDesign and evolve the agent runtime, the core loop handling reasoning, tool use, planning, memory retrieval, and response generationBuild and ship agent capabilities across modalities (e.g. image/video, voice, browsing, code execution)Own LLM orchestration and model routing across multiple providers, optimising latency, cost, reliability, and qualityImplement memory systems that allow agents to learn from interactions (long-term memory, episodic recall, semantic retrieval)Prototype and productionize autonomous behaviours such as proactive task execution, scheduling, and goal-directed workflowsCreate evaluation frameworks and metrics that measure agent quality, personality consistency, and real user impactWhat Great Looks LikeYou’ve personally built and shipped agentic systems, not just prompt wrappers or demosYou’re comfortable owning ambiguous, greenfield problems and turning ideas into working product fastYou think in systems: distributed workflows, multi-step reasoning, orchestration, reliabilityYou code daily and care deeply about performance, UX feel, and real-world usefulnessWhy JoinJoin at a genuine product inflection point, early access launch, new architecture direction, and strong internal momentumWork in a small, elite engineering cohort where each senior hire has outsized ownership and influenceHelp define the company’s next-generation agent platform and model infrastructure from the ground upCollaborate closely with product leadership and shape how consumer AI agents evolve in the real worldClear trajectory toward technical leadership and founding-level impact as the organisation scalesIf you’ve built real agent systems and want to work on problems that don’t have playbooks yet, please apply with your resume!

Staff Agentic AI Engineer / AI Engineer $200,000 - $270,000 San Francisco, CA Onsite, 5x per week in office Full time / Permanent Most "agentic AI" roles are still research projects dressed up as products.This one isn't. The agent is live, in production, handling real workflows for Tesla, BMW, Meta, and Amazon. You’ll be in the codebase from day one, improving an agent that's already shipping, not starting from a blank page.This company builds AI-powered automation for mechanical engineers - Claude Code for CAD, CAE, and PLM workflows. $32M raised, backed by Eric Schmidt and early investors in Anthropic and OpenAI. Over 1,000 customers running production workflows on the platform.They're hiring a Staff Engineer to own the agent intelligence layer, the system that takes an engineer's intent and executes it reliably across complex desktop software. What you'd own:Evals infrastructure and agent benchmarking - defining what "good" looks like for a domain with no established benchmarkAgent harness build-out and ongoing performance improvement (task success rate, token efficiency, workflow coverage)Architecture decisions: tool-calling strategy, state management, context handling, model routing, error recoveryTechnical leadership of a small team of AI engineers - player-coach, not pure manager The profile they're looking for:Proven expertise building agentic systems into production. Systems that take real actions - tool-calling, multi-step state, failure handling, cost constraints. Ideally 2 yearsStrong on evals: task completion, failure mode analysis, regression detectionProduction-first mindset - you'd rather ship 70% coverage reliably than demo a clever system you can't measureBuilders over researchers. Too much academic background is a flag here Desktop automation experience is nice-to-have. Mechanical engineering background is nice-to-have. Experience taking agentic systems into production is the bar.

Machine Learning Engineer – Speech Model Training $250,000 - $300,000 San Francisco, CA Hybrid, 3x per week in office Full time / Permanent In this role you won’t be wrapping APIs or fine-tuning existing models. You’ll be building models across raw acoustic signal processing all the way through to production inference on edge devices. At a company that actually ships to 1.5M live users. A profitable, fast-growing AI company ($250M ARR in under three years, no VC dependency) is standing up a SpeechLLM lab from scratch. This is a founding seat on that team. They build a hardware-software AI companion used daily by over 1.5 million professionals worldwide. The next chapter is a world-class speech intelligence core and they need the engineers to architect it. What you'd own:Design and train large-scale speech models end-to-end. Unified SpeechLLMs, ASR, expressive TTS, generative audioOwn the full stack from acoustic feature engineering to GPU cluster optimisationRun and optimise distributed training at scale via PyTorch or JAX, FSDP, DeepSpeed, etcDrive real-time inference performance with vLLM, TensorRT-LLM, or SGLangApply RL alignment techniques to improve conversational qualityDebug the hard problems in distributed infrastructure and ship solutions You likely have:Proven experience training large-scale audio or speech models from the ground upDeep PyTorch or JAX expertise with real distributed training experienceGenuine comfort traversing the entire ML stack from signal processing to productionA bias toward shipping: you take ownership, you iterate fastStrong bonus: neural audio codecs, diffusion/flow-matching architectures, or LLM pretraining experience. Why joinProfitable company at ~$250M run rate - you'll see the impact of your work immediately in a product used daily by professionals worldwideDirect ownership of the live speech quality stack, not a supporting role in a large orgHybrid San Francisco team with real access to large, diverse, multilingual audio datasetsShort feedback loops - improvements ship fast and metrics are visibleClear path toward senior technical leadership as the audio team grows