Job Requirements
About Phenom and the Agents team
Phenom builds the AI talent platform that large enterprises use to hire, onboard and grow their people. The Agents team builds the AI agents that do hiring work end to end: the Sourcing Agent, the Intake Agent, outreach and candidate-fraud detection, along with the platform products that power them.
Our bet is goal-driven hiring. A hiring manager states an outcome, such as "20 qualified ICU nurses in Dallas this quarter", and agents work backwards through intake, sourcing and outreach to deliver it. These agents run inside Phenom and inside the tools recruiters and managers already use, connected through open standards like MCP and agent-to-agent protocols.
The work is high-stakes: agents act on candidate data and influence employment decisions for some of the largest employers in healthcare, retail and financial services.
About the role
You will own a portfolio of products end to end: what they do, how well they do it, and whether customers get the hiring outcomes they paid for. The role is one part platform product manager and one part forward-deployed builder. You will define how our agents are designed, evaluated and governed, and you will also sit with enterprise customers to map their hiring workflows, demo working agents, and see deployments through to steady state.
You may also own the platform products these agents depend on, such as the hiring ontology, sourcing and data collection, document intelligence, and the hiring-data pipelines that feed reinforcement learning. A great agent is only as good as the data, structure and feedback loops beneath it.
We want someone who builds, not someone who only manages a roadmap. You prototype with coding agents before writing a spec, you write your own evals, and you can explain exactly why an agent failed on a real customer's data.
Location: Hyderabad, India, working closely with customers worldwide, sales and leadership
What you'll do
Own an agent
Own the vision, roadmap and outcomes for an agent (for example, sourcing, intake or fraud), from problem definition through launch, adoption and iteration.
Define each agent's job in outcome terms: what goal it accepts, what it is allowed to do on its own, and when it hands back to a recruiter or hiring manager.
Own the metrics that prove agents work, such as goal-fulfilment rate, qualified candidates delivered, time to first qualified slate and recruiter acceptance, and report them in weekly business reviews.
Own the platform products that power agents
Own one or more shared platform products that agents build on, for example:
Hiring ontology: the structured model of skills, roles, credentials and requirements that agents reason over.
Sourcing and data collection: how candidate and job data is acquired, enriched, deduplicated and kept fresh, with consent and provenance tracked.
Document intelligence: parsing and understanding resumes, job descriptions, credentials and onboarding documents.
Hiring-data RL pipelines: turning recruiter decisions, candidate responses and hiring outcomes into governed training and reinforcement-learning signals that make agents better over time.
Treat each platform product as a product: define its internal and external customers, roadmap, quality metrics and service levels.
Decide what belongs in the shared platform and what belongs inside a single agent, so new agents reuse capabilities instead of rebuilding them.
Set the quality bar with evals
Build and maintain evaluation sets drawn from real customer workflows, and decide what "good enough to ship" means for each agent.
Run regression, LLM-as-judge and human-review loops with engineering and data science, and turn failures into prioritised fixes.
Balance accuracy, latency and cost as models and prompts change.
Work forward-deployed with strategic customers
Lead discovery with enterprise talent-acquisition teams: map their hiring workflows, systems and constraints, and find where agents create the most…