Job SummaryPantomath is hiring an Applied AI Engineer, Agents to build AI agents that solve complex enterprise data problems. You'll combine applied research with hands-on engineering to develop new capabilities, evaluate what works, and bring promising approaches into production. The role spans models, agent systems, and product development: you'll explore emerging techniques, run rigorous experiments, and own the software that turns research into useful customer experiences. You'll have meaningful ownership of both the research and the engineering, with a direct path from an idea to a capability customers use.
What You'll Do- Build and ship AI agents. Develop agentic products that solve complex enterprise problems and deliver meaningful customer value.
- Advance model and agent capabilities. Explore fine-tuning, post-training, and other adaptation techniques to improve performance on enterprise tasks.
- Design experiments and evaluations. Build datasets, benchmarks, and feedback loops to measure progress and understand failure patterns.
- Build the systems behind the agents. Develop services, integrations, and infrastructure that support reliable execution at scale.
- Bring research into production. Turn promising experiments into product capabilities, owning implementation, deployment, and ongoing improvement.
- Improve production performance. Make practical tradeoffs across quality, reliability, latency, cost, and security.
- Shape the product and research direction. Partner with engineering, product, and customer-facing teams to identify valuable problems and pursue approaches that work.
What You'll Need- Experience building and shipping AI systems, with substantial hands-on contributions to code, experimentation, and production engineering.
- Hands-on experience with post-training tools such as PyTorch, Hugging Face Transformers, TRL, and PEFT, including supervised fine-tuning, parameter-efficient adaptation such as LoRA, and preference optimization or reinforcement learning.
- Practical experience with LLMs, agent systems, and model evaluation.
- Experience adapting or training models, with an understanding of data quality, experimental design, and generalization.
- Exposure to training and evaluation data pipelines, synthetic data generation, or human feedback systems.
- A working knowledge of agent orchestration, tool use, retrieval, or inference optimization.
- Comfort working with enterprise data platforms, distributed systems, or production AI infrastructure.
- Familiarity with distributed GPU training and efficient inference using tools such as FSDP, DeepSpeed, or vLLM, alongside reproducible experimentation and model evaluation.
What You'll Bring- Strong software engineering fundamentals and proficiency in Python, TypeScript, or comparable languages.
- The ability to read research, reproduce useful results, and assess whether an approach will translate to real-world performance.
- Comfort owning ambiguous problems and moving between research exploration and product delivery.
- Clear communication, technical judgment, and a collaborative approach.
Pay Range for this role is 150,000 - 230,000 
