About the Role
We're looking for a Senior Agent Engineer to push the boundaries of what AI agents can do at Pika. You'll work on the systems that give AI agents their personality, memory, reasoning, multi-modal capabilities, and ability to act autonomously across platforms. This is the core of what makes Pika's AI products feel alive.
You'll design agent architectures, build tool-use frameworks, optimize LLM interactions, and create the systems that allow agents to learn, remember, and evolve over time. If you're excited about building AI systems that feel genuinely intelligent — not just wrappers over chat APIs — this role is for you.
What You'll Do
- Design and evolve the agent runtime — the core loop that handles reasoning, tool use, memory retrieval, and response generation
- Build agent capabilities — image generation, voice synthesis, video creation, web browsing, code execution, and other skills
- Optimize LLM integration — prompt engineering, context window management, multi-provider model routing (Claude, GPT, Gemini, open-source), cost optimization, and latency reduction
- Implement memory systems — long-term memory, working memory, episodic recall, and semantic search so agents learn from every interaction
- Design autonomous behaviors — proactive actions, scheduled tasks, and goal-directed behavior that makes agents feel self-directed
- Build evaluation and quality systems — benchmarks, A/B testing, and metrics for agent behavior, personality consistency, and response quality
- Experiment with new architectures — multi-agent collaboration, planning, chain-of-thought reasoning, and other emerging patterns
- Collaborate with product and design to translate AI capabilities into intuitive user experiences
What We're Looking For
- 5+ years of software engineering experience, with 2+ years working with LLMs or AI agent systems
- Deep understanding of LLM capabilities and limitations — you know how to get the best out of frontier models
- Experience building agent systems — tool use, function calling, multi-step reasoning, retrieval-augmented generation (RAG)
- Strong prompt engineering skills — system prompts, few-shot learning, chain-of-thought, structured output
- Proficiency in TypeScript and/or Python
- Understanding of embedding models and vector search for memory and retrieval
- Comfortable with rapid experimentation — you ship experiments, measure results, and iterate
- Product intuition — you understand what makes an AI agent feel "alive" vs. robotic
- Clear communication skills and a team-first mindset
Nice to Have
- Experience with multi-modal AI (image generation, TTS, speech-to-text, video generation)
- Experience with agent frameworks (LangChain, AutoGPT, CrewAI, or custom runtimes)
- Background in NLP, computational linguistics, or cognitive science
- Experience with fine-tuning, RLHF, or DPO
- Familiarity with AI safety and alignment considerations
- Experience with real-time/streaming LLM responses
- Previous startup experience — comfortable with ambiguity and moving fast