About this role
Traackr is a global SaaS technology company providing a data-driven influencer marketing platform that marketers use to optimize investments, streamline campaigns, and scale programs. Our customers range from some of the world’s largest companies in the beauty and personal care space to digitally native indie brands, which have all made influencer management and engagement a critical practice of their marketing and advertising programs. We are a remote-first company, and for the folks that like to meet in person, we have offices in San Francisco, New York, Boston, Paris, and London. We operate on a culture of mutual respect, with core value pillars including:
Trust. We earn the trust of our team, customers, creators, and partners through transparency, predictability, and integrity.
Diversity. Bringing diverse perspectives to the table results in stronger outcomes. All are welcome.
Value. Through our words and actions, we strive to create tangible value for our customers and peers. We only succeed when our community succeeds.
Ownership. We lead with action. We take pride in solving the hardest challenges and feel accountable for our commitments.
Mutual success. We share goals with each other and with our clients. Alignment, collaboration, and empathy are the cornerstones of our success.
This position is 100% remote, with the understanding that occasional in-person attendance may be required for trainings, meetings, and team gatherings, as determined by your manager.
As a Senior Data Scientist, you will help push Traackr’s product and business forward by turning ambiguous customer problems into measurable outcomes. You’ll partner closely with Product and Engineering to design experiments, develop and evaluate ML, AI, Statistics, and recommendation capabilities (from classic ML models, to LLM-powered automations, to large scale search capabilities). As a Data Scientist at Traackr you will establish reliable practices for evaluation, monitoring, and continuous iteration. You’ll also raise the bar across teams by coaching others on experimentation and AI evaluation best practices.
\nResponsibilities- Work with Product and Engineering from intake through discovery: turn a vague customer problem into a defined question, explicit acceptance criteria, and success metrics that can be measured after launch.
- Make the build-versus-buy call and write down the reasoning. Our default is to use frontier LLMs the way we use cloud infrastructure. Custom models need a real argument behind them: measurably better performance, a proprietary data advantage, a compliance constraint, or economics that hold at our volume.
- Work up the ladder of complexity, not down it. A query or heuristic before a prompt, a prompt before retrieval, retrieval before fine-tuning, fine-tuning before a custom model. Part of the job is knowing when the simple option genuinely will not clear the bar, and being able to show it.
- Revisit existing models with the same scrutiny. Some of what we built when custom modeling was the only option is now worth re-examining.
- Build prototypes fast against real data, in the team's repo, following the team's conventions. Label the shortcuts. A notebook nobody else can run is not a deliverable.
- Support the engineers and data teams who own the production system: define data requirements, validate behavior, help debug quality problems in hypercare, and stay involved after launch.
- Know where the boundary between experimentation and engineering sits, and be comfortable when your work crosses it and someone else takes over.
- Build the evaluation layer for both ML and LLM features: golden datasets, offline and online evaluation plans, regression suites that run in CI on the pull requests that touch AI code, and monitoring that catches quality drift before a customer does.
- Do real error analysis. True and false positives and negatives, by segment, with a view on which failures customers will actually care about.
- Run experiments where the product supports it: hypothesis, guardrail metrics, A/B or quasi-experimental design, and a readout that ends in a decision rather than a deck.
- Design and evaluate the agent harnesses behind our core agentic journeys: task decomposition, tool design, context management, fallbacks, observability, and cost and latency budgets.
- Share with others where an agent adds value and where a deterministic pipeline is the better product. Some of the most useful work here will be talking a team out of an agent.
- Coach engineers and PMs on prompt patterns, tool and function calling, structured outputs, guardrails, and how to tell whether a change actually improved anything.
- Build reusable assets: evaluation harnesses, shared datasets, templates, documentation, and the occasional workshop.
- Communicate tradeoffs clearly to technical and non-technical audiences. A lot of your impact will come from a well-argued written recommendation.
- Handle data responsibly: appropriate data handling, bias and fairness considerations where they apply, and human-in-the-loop workflows where the cost of being wrong is.
- 4+ years (or equivalent) of data science or applied AI work that shipped to production users or materially changed product direction.
- At least one LLM-powered feature shipped to real users, not a prototype or internal demo. We will ask what broke after launch and what you changed.
- Strong Python and SQL, with the ability to write maintainable, reviewed, tested code in a shared repository.
- Range across the toolkit: heuristics, statistics, classical ML, and LLM-based approaches, with enough command of each to argue for the right one and against the wrong one.
- Evaluation discipline: dataset design, error analysis, offline and online measurement, regression testing, and monitoring for drift.
- A track record of working cross-functionally with Product and Engineering, and of influencing decisions without owning the team.
- Evidence of making other people better at this, through mentorship, enablement, documentation, or teaching.
- Strong applied statistics and experimentation skills (A/B testing, causal thinking, metric design, interpretation under uncertainty)
- Proven ability to evaluate and improve models in real conditions: dataset design, error analysis, offline metrics, online measurement, monitoring, and iteration
- Hands-on experience building with LLMs in product contexts, including some of: RAG/grounding, tool/function calling, structured outputs, prompt iteration, quality/cost/latency tradeoffs
- Practical approach to LLM evaluation: golden sets, regression testing, human review loops, and monitoring for quality drift
- Experience with modern MLOps/LLMOps practices (experiment tracking, ETL pipelines, versioning, CI/CD for ML, observability)
- NLP and information extraction/classification on noisy social/content data
- Experience developing and evaluating large scale Retrieval, Recommendation- and Search Systems.
- Multi-step or agentic system design: tool use, orchestration, context management, failure handling, cost and latency optimization.
- Retrieval, search, ranking or recommendation systems at scale, including how to evaluate them.
- Applied statistics and experimentation: causal thinking, metric design, power analysis, interpretation under uncertainty.
- NLP, information extraction and classification on noisy social and creator content.
- MLOps and LLMOps: experiment tracking, versioning, CI/CD for models and prompts, observability.
- Classical ML in production, including the maintenance and retraining burden that comes with it.
Benefits
- Competitive Salary
- Remote Work Options with Hybrid Flexibility and Home Office Set-Up Stipend
- Coworking Office Subscription for Collaborative Spaces
- Health, Dental, and Life Insurance Coverage*
- Open Vacation Policy and Flexible Holiday Schedule to Suit Your Needs
- Paid Parental Leave to Support Quality Time with Your Loved Ones
- Career Development, including Internal and External Training Opportunities
Traackr employs individuals in multiple US states and countries. We use market benchmark data and geographic zones to determine our salary ranges. Your zone's specific pay range is dependent on your home location. We encourage you to discuss your zone-specific pay range with your Traackr recruiter for more details.
Benefit programs vary by country/state of residency, are subject to eligibility requirements, and may be modified from time to time. Ask for more details about the benefits in your specific region.
Traackr is an Equal Employment Opportunity employer. We do not discriminate based on race, religion, color, national origin, gender, sexual orientation, gender identity, gender expression, age, status as a protected veteran, status as an individual with a disability, or other legally protected characteristics. All your information will be kept confidential in accordance with EEO guidelines.
Unsolicited resumes
Traackr does not accept unsolicited resumes/CVs from headhunters or recruiting agencies sent directly to Traackr employees or through our website. Traackr will not pay any fees to any third-party agency or company unless there is a signed agreement with Traackr.
Privacy
Traackr, Inc. has published a Privacy Notice, including CCAP for California and GDPR policies for its UK and European Union subsidiaries, accessible at https://www.traackr.com/privacy-policy.
All questions, comments, and requests regarding data processing at Traackr should be addressed to [email protected].
Tired of cold applications?
Sign up with Clera and we'll reach out the moment a role actually fits you — no more spraying applications into the void.
Know someone who'd be great for this?