Senior Director, Technical Product Management — AI Quality & Evaluation
https://www.salesforce.com
Job Details
Job Type: Full-time
Work Mode: Remote
Location: San Francisco Metro, California or Seattle, Washington
Salary:
$218,400 – $365,200 base annually; $263,200 – $401,400 in SF Bay Area (base only, excludes bonus, equity, benefits) per year
Job Description
About the Role
Salesforce's Agentforce incubation team operates with the speed and ownership of a startup while drawing on the distribution and customer base of one of the largest enterprise software companies in the world. This team is now hiring a Senior Director, Technical Product Management to own how Salesforce defines, measures, and improves the quality of its AI agents — turning frontier model capability into something enterprise customers can actually trust in production.
This is a hands-on, individual-contributor leadership role (through at least 2027) rather than a people-management position, though coaching responsibilities — especially of junior designers and technical program managers — come built in. You'll sit at the intersection of applied AI research, engineering, and product, working daily with researchers and applied scientists to close the gap between "the model behaves well in a demo" and "the model behaves well for every customer, every time."
Key Responsibilities
- Shape the overall strategy and roadmap for AI quality and evaluation, tying together what customers need, what's technically achievable, and what actually moves the business.
- Work with research, engineering, and product partners to agree on what "good" looks like — shared quality bars, evaluation coverage, and the criteria that decide whether something is ready to ship.
- Make the hard calls on where investment goes: customer-facing quality fixes, evaluation infrastructure, or new product features — and get the team aligned behind those calls.
- Personally dig into why an agent got something wrong — a bad answer, a failed multi-step workflow — define what correct behavior should look like, and work with engineers and researchers to test and confirm a fix actually helps real customers.
- Build a working loop that connects customer complaints, evaluation data, live experiments, and shipped releases, so the team keeps learning instead of guessing.
- Apply real ML fundamentals — training/test separation, overfitting risk, representative coverage, independent validation — when judging whether an evaluation or a grader can actually be trusted.
- Prototype customer-facing tools that turn a failed agent interaction into something concrete — a documented evaluation case a customer can use to confirm a fix really worked.
- Mentor and raise the bar for product partners, junior designers, and technical program managers without needing formal reporting authority over them.
- Translate technical nuance — including where Salesforce's AI genuinely is and isn't ahead of competitors — into plain language for customers and senior leadership.
What You'll Need
- A track record of shipping AI/ML products with real, measurable customer outcomes, taking on steadily more ownership along the way.
- Direct, hands-on experience building AI agents and the evaluations that test them, working shoulder-to-shoulder with researchers or applied scientists — even 1–2 focused years in generative AI can be enough if your fundamentals and track record are strong.
- Sharp applied judgment on model behavior, prompting, retrieval, and agent/tool workflows, including the tradeoffs between fixing the model, the data, the system, or the product.
- Comfort getting into the weeds yourself: reading raw traces, prototyping with coding agents, and reasoning rigorously about overfitting, edge cases, and uncertainty in a system that behaves non-deterministically.
- A flexible, first-principles way of thinking — you can frame the real underlying problem, question your own assumptions, and change course when the evidence says you should.
- Experience setting shared standards across teams and resolving disagreements without losing momentum or clarity on who owns what.
- A demonstrated habit of coaching and mentoring colleagues even without formal authority over them.
- Communication skills sharp enough to tell customers and executives, plainly, what's proven, what's promising, and what's still unknown.
Nice to Have
- Background in evaluation platforms, agent systems, AI observability, retrieval, or other enterprise AI products.
- Experience inside an AI-native startup or similarly entrepreneurial setting, carrying a product from first customer signal through to real adoption.
- A habit of prototyping with coding agents and shipping customer-facing performance or outcome-driven experiences yourself.
- Experience pulling research, engineering, product, and customer-facing teams into agreement on a shared plan.
If you'd rather be elbow-deep in agent traces than managing a Gantt chart, and you want your evaluation work to directly decide what millions of Agentforce users experience, this is built for you.
Job Benefits
Salesforce bundles its benefits under a program called B-Well, built around covering fundamentals, staying inspired, and building habits that last. Highlights include:
- Comprehensive health coverage — medical, dental, and vision plans.
- Mental health support for everyday wellbeing, not just crisis moments.
- Generous time-off programs built around genuine rest and recovery.
- Paid parental leave, plus life and disability insurance.
- 401(k) retirement plan and an employee stock purchase program (ESPP).
- A wellbeing reimbursement employees can put toward things like massage therapy or other local wellness services.
- Virtual musculoskeletal/physical-therapy support (via a partner like Hinge Health) for recovering from injury without an in-person clinic.
- Life-stage support, including parenting resources (via a partner like Cleo) for employees navigating early-parenthood milestones.
- Consideration for qualified applicants with arrest and conviction records, per San Francisco and Los Angeles Fair Chance hiring ordinances.
This listing is from the company career pages and for more details follow the link from the Apply button.
Apply Now
This job is posted by an independent employer. Placeman does not verify employer identity or guarantee responses. We encourage you to research the company and role before applying. Questions? Email us at
support@placeman.co.in