At a glance
- Employment
- Full-time
- Location
- San Francisco Bay Area
About GT HQ
- Ema builds AI Employees for HR, IT and Finance. Our AI Employees take on the busy work across the employee experience, from recruiting, onboarding and benefits to IT support, invoice processing and payroll, so people can spend their time on work that needs them. Founded by former executives from Google, Coinbase, Flipkart, and Okta, our team includes engineers from premier tech companies and graduates of Stanford, MIT, UC Berkeley, CMU, and IITs.
- We are backed by industry leading investors including Accel, Naspers/Prosus, Creaegis, Section32, and angels like Sheryl Sandberg and Dustin Moskovitz. Headquartered in Silicon Valley and with offices in London, Bangalore and Vancouver, Ema is at the frontier of what Agentic AI can do in production. We ship real systems that run real business processes at scale.
- BUILD AGENTS THAT LEARN FROM THE WORK THEY DO.
- Ema builds AI employees that carry out complex workflows across enterprise applications. Our ML team works on the loop that makes them better: production traces become data, data becomes training and evaluation, and better agents produce better traces. The hard part is deciding which intervention will improve behavior in the next real workflow.
About the role
- Ema is seeking a Software Engineer to build and improve AI Employees that carry out complex workflows across enterprise applications.
- The role involves combining applied research with production engineering to develop agentic AI systems that learn from their own performance.
- Engineers will own the full lifecycle of projects, from defining problems and running experiments to deployment and monitoring.
Responsibilities
- Design context, tools, skills, and orchestration for multi-step agents.
- Build self-improvement loops with explicit permissions, evaluation gates, and rollback.
- Curate trajectories for SFT, optimize preferences, or run RL on real agent tasks.
- Turn enterprise workflows into reproducible training and evaluation environments.
- Mine production agent-steps for failures and build curated corpora.
- Calibrate judges against human labels and construct behavior-level benchmarks.
- Connect enterprise information with user- and tenant-level learnings.
- Build and evaluate routing, ensembles, caching, and small-model specialization.
- Define problems, establish baselines, and conduct experiments.
- Manage serving and integration, following systems through to production handoff.
Requirements
- Substantial work in at least one of: agent/tool-use systems, post-training, reward modeling, RL environments, retrieval and memory, or evaluation design.
- Evidence of zero-to-one ownership of a system, model, or research project.
- Ability to size experiments, choose baselines, and account for variation across tasks.
- Ability to debug across model and system boundaries to produce maintainable code.
- A master’s or PhD in a relevant field, or equivalent work or research experience.
Preferred qualifications
- Open-model post-training with TRL, veRL, OpenRLHF, or custom loops.
- Experience debugging reward hacking or unstable optimization.
- Experience with interactive agent environments or harnesses for software engineering, web, or tool use.
- Experience with large-scale trace analysis, data curation, or synthetic generation.
- Experience designing systems for complex, long-horizon agent work across multiple modalities.
- Experience with serving using vLLM or SGLang, distillation, quantization, or multi-node GPU training.
- Practical security work on prompt injection, data governance, or permission boundaries.
Benefits
- Variable compensation
- Equity
Skills
- DPO
- GRPO
- DAPO
- TRL
- veRL
- OpenRLHF
- vLLM
- SGLang
