Machine Learning Engineer Graduate (Conversational AI)

Build a self-evolving unified conversational agent for TikTok’s global e-Commerce customer service in Seattle.

Responsibilities

• Agent runtime (harness / agent loop): orchestrate skills, tools, and context; implement loop control and intervention; apply progressive disclosure; add behavior-level guardrails
• Production safety layer: create pre-flight budgets and timeout truncation; implement serve-time gates; deliver shadow and swap-in answers; define safe fallback paths
• Context & memory for long multi-turn agents: build agentic memory via structured note-taking; perform context compaction and summarization; support context editing and observation masking; use retrieve-then-load for just-in-time retrieval
• Post-training and the data flywheel: apply SFT, DPO, and RL so rules move into model weights (reducing prompt length); distill to smaller serving models; convert served conversations into training, evaluation, and retrieval signals
• Tools, skills, and MCP: design tools as APIs and connectors; enable skill and tool search for large inventories; govern a skill library (handle description conflicts, trigger evaluations, maintain cross-skill mis-fire matrices); load skills on demand instead of placing every definition into context
• Evaluation you can launch with: use LLM-as-judge with human-agreement calibration; apply statistical rigor (paired comparison, confidence intervals, repeated sampling, pass^k); run held-out and time-rolling evaluation splits with overfitting alarms; support cascaded scoring and cross-family judge panels to scale evaluation cost
• Self-evolving loop: implement case mining to derive root cause; generate constrained candidate sets; replay-verify against frozen regression sets; run canary checks and feed results into the flywheel
• Auditable systems: build candidate registry with exact runtime read-back, change lineage, and an archive of rejected candidates for sampling in later rounds
• Online experimentation and causal readout: run shadow, canary, and A/B experiments; use non-inferiority gates and traffic-split health; define metrics that hold up under scrutiny; apply off-policy counterfactual evaluation when live A/B is not possible

Requirements

• Completing or recently completing a Bachelor’s degree in Computer Science, Electrical Engineering, Mathematics, Statistics, or a related discipline
• Strong Python plus one of: C++, Go, Rust, Java
• Solid fundamentals in ML/DL/NLP with hands-on experience using LLMs or agents (coursework, research, internship, competition, open-source, or a serious side project)
• Basic statistical literacy: confidence intervals, p-values interpretation, and understanding the difference between “a number went up” and real system improvement
• Ability to read a research paper or engineering blog and translate it into working code

Technology

• Python, C++, Go, Rust, Java
• SFT, DPO, RLHF, RLAIF, RLVR
• vLLM, TensorRT-LLM, MoE, KV
• MCP, A/B, LLM-as-judge
• ACM-ICPC, Kaggle

Benefits

• Day one access to medical, dental, and vision insurance
• 401(k) savings plan with company match
• Paid parental leave
• Short-term and long-term disability coverage
• Life insurance
• Wellbeing benefits
• 10 paid holidays per year
• 10 paid sick days per year
• 17 days of Paid Personal Time (prorated upon hire with increasing accruals by tenure)

Team introduction

• Develop the next-generation unified agent system for TikTok global e-Commerce customer service across 30+ languages
• North star is a self-evolving agent: post-training, harness, memory/context engineering, tools, and evaluation form a closed loop where served conversations become training, evaluation, retrieval, and skill-induction signals
• Evaluation and experimentation are treated as first-class systems
• Build the agent runtime itself, not prompts on top of a vendor API
• Work spans generative recommendation, large recommendation models, multimodal representation learning, and cross-domain value modeling

Preferred qualification(s)

• Experience building the runtime rather than only calling an API: agent loop/harness, memory/context management, RAG or tool-use agent, or fine-tuned and post-trained models (research/hobby/competition scale included)
• Hands-on depth in at least one area:
  • Post-training: SFT, DPO, RLHF, RLAIF, RLVR; reward modeling; reward hacking and defenses
  • Agent systems: harness, context engineering, MCP/skills, sub-agents, tool search
  • Evaluation & experimentation: LLM-as-judge and judge calibration, pass^k, regression suites, A/B and non-inferiority testing, off-policy evaluation
  • Self-improving/evolutionary systems: evolutionary program search, candidate archives and parent sampling, automatic prompt/context optimization, multi-objective (Pareto) selection, credit assignment
  • Inference & serving: vLLM/TensorRT-LLM, MoE, KV/prompt caching, cost engineering
  • Multilingual NLP
  • Publications and achievements: PhD publications, strong ACM-ICPC/Kaggle/ML competition results, or notable open-source contributions
  • Result honesty: preference for acknowledging when results are not significant
• e-Commerce or multilingual experience is a plus, not required

Other application constraints

• Successful candidates must commit to an onboarding date by the end of the year
• Please state availability and graduation date clearly in the resume
• Candidates can apply to a maximum of two positions; applications are reviewed in the order submitted
• Applications reviewed on a rolling basis; apply early

Compensation & pay transparency

• Base salary range (selected city): $121,600 - $243,200 annually
• Compensation may vary based on qualifications, skills, competencies, experience, and location
• Base pay is part of the total package; role may be eligible for additional discretionary bonuses/incentives and restricted stock units

Location & employment details

• Location: Seattle, WA (onsite)
• Employment Type: Regular
• Job Code: A31120
• Start: 2027

Accommodation

• TikTok provides reasonable accommodations in recruitment processes for candidates with disabilities, pregnancy, sincerely held religious beliefs, or other protected reasons
• Contact: https://tinyurl.com/RA-request

Los Angeles County (Unincorporated) candidates

• Qualified applicants with arrest or conviction records will be considered in accordance with the Los Angeles County Fair Chance Ordinance for Employers and the California Fair Chance Act
• Company notes criminal history may relate negatively to job duties, potentially resulting in withdrawal of a conditional offer