Senior Director, Frontier AI Research, Bay Area, NY

EPAM Systems•San Francisco, CA

About The Position

EPAM's new Frontier AI business unit partners directly with leading AI labs and advanced AI organizations, translating their research and post-training objectives into technically rigorous, deliverable programs across evaluations, RL environments, and training data. The Senior Director, Frontier AI Research is EPAM's first dedicated Frontier AI research hire, operating across a broad set of capabilities rather than a single research domain. The role sits at the intersection of research, solution architecture, and technical sales, remaining technically engaged through early pilots while delivery leadership owns day-to-day execution. The right person is a genuine technical peer to sophisticated AI researchers: someone who can translate ambiguous research goals into concrete specifications that delivery teams can execute at quality and scale.

Requirements

Significant experience in machine learning, AI research, or research engineering, with demonstrated work on modern deep learning and large language models
Strong understanding of the model training and post-training lifecycle, including SFT, RLHF, reward modeling, and evaluation
Demonstrated depth in one or more Frontier AI focus areas (evaluations, RL environments, agentic systems, reward or verifier design, coding agents), with the breadth to operate across adjacent areas
Experience translating ambiguous research objectives into structured experiments, datasets, evaluations, or environments
Hands-on technical capability, including strong Python skills and experience with modern ML frameworks, LLM APIs, and evaluation tooling
Demonstrated ability to communicate complex technical concepts clearly to both technical and cross-functional audiences
Advanced degree in Computer Science, Machine Learning, AI, Statistics, or a related field, or equivalent demonstrated research experience

Nice To Haves

Peer-reviewed AI/ML research or a strong record of technically substantive applied research
Prior experience building solutions or conducting research for frontier AI labs or leading foundation model companies
Direct experience with RLHF, RLAIF, preference optimization, or synthetic data generation and curation
Comfortable operating as a genuine technical peer to sophisticated researchers, with the commercial awareness to translate research credibility into customer trust
Entrepreneurial mindset, comfortable building a new capability and creating structure where established process does not yet exist

Responsibilities

Engage directly with researchers, engineers, and technical leaders at frontier AI labs to understand model training, post-training, and evaluation objectives
Translate ambiguous research goals into explicit target capabilities, failure modes, and success criteria, advising customers and internal teams on the right technical approach
Design rigorous evaluation approaches, including task taxonomies, benchmarks, rubrics, and graders, and analyze model outputs to identify failure modes and improvement opportunities
Help architect RL environments, reward functions, and verifiers that provide reliable training and evaluation signals while mitigating reward hacking and weak verification
Design technically rigorous training data programs (SFT, preference data, RLHF), defining task distributions, annotation criteria, and quality standards
Translate recurring customer research needs into repeatable EPAM Frontier AI offerings, methodologies, and technical assets
Partner with the Platform and Operations Lead to determine the tooling, infrastructure, and specialist talent required to deliver new offerings
Serve as the primary research and technical SME supporting Frontier AI Sellers in strategic customer pursuits, from discovery through solution design
Translate customer requirements into technically compelling solution designs, proposals, and pilot plans that establish credibility with sophisticated AI research organizations
Convert research requirements into clear specifications, including scope, rubrics, and quality thresholds, that delivery teams can execute without losing the research intent
Provide technical oversight during early and strategically important engagements, reviewing outputs and model behavior to confirm the program is producing the intended results
Maintain strong familiarity with developments in frontier model training, post-training, RL, agentic systems, and evaluation, and apply that knowledge to EPAM's offerings
Build relationships across the Frontier AI research ecosystem and contribute to technically credible customer facing content and thought leadership