Lead Data Scientist Applied AI - USA

Socket.dev; Santa Clara (CA); On-site; USD 150,000 - 170,000; Full time

Benefits offered by this job

Unlimited PTO

Generous parental leave

Annual bonus program

Employer Stock Purchase Program

Medical, dental and vision coverage

Mental health and wellbeing support

Job summary

Cogniify is seeking a Lead Data Scientist to build advanced AI systems, measure performance and uncertainty, and explain results for decision-makers. You’ll work across Generative AI, computer vision, forecasting and optimization with a focus on production-grade reliability and business impact.

You will drive experimental design, error analysis and deployment, ensuring models meet production readiness and governance standards across multi-step, tool-using and multi-agent workflows.

Qualifications

• Justify modeling decisions with evidence and hypotheses.
• Quantify risk, uncertainty, bias and drift.
• Explain model behavior using SHAP/LIME and dashboards.
• Translate results into business impact and ROI.

Responsibilities

• Design problem framing, data strategy and deployment for ML/LLM/agentive systems.
• Build production-ready solutions for unstructured data (image, text, video, time-series).
• Deliver computer vision components with measurable benchmarks (detection, classification, tracking).
• Develop forecasting models and integrate into operational workflows.

Skills

Structured experiments

Explainability techniques

Model evaluation

Business impact translation

Strong communication

Production AI deployment

Education

Bachelor's or Master's in CS/DS/Statistics

Tools

Docker

Kubernetes

PyTorch

LangGraph

LangChain

Job description

The Role

We are seeking a Lead Data Scientist who can build advanced AI systems and demonstrate, with evidence, why they are accurate, reliable and appropriate for production. This role combines strong statistical and machine learning expertise with hands‑on experience designing and delivering production AI solutions.

You will work across Generative AI, Agentic AI, computer vision, forecasting and optimization. Your central responsibility will be to measure model performance and uncertainty, explain model behavior, identify risks and failure modes, and connect technical results to business outcomes that leaders can use to make decisions.

The ideal candidate is comfortable running structured experiments, presenting error analysis to technical and non‑technical stakeholders, and taking models from problem definition through deployment and monitoring. You should know when Generative AI is the right solution, when a simpler statistical approach is more effective, and how to support that decision with data.

What You Will Do

Scientific Rigor and Model Evaluation

• Justify modeling decisions with evidence: Frame business problems, form hypotheses, run structured experiments and select between classical machine learning, deep learning and Generative AI using measured performance, cost and risk.
• Quantify risk and uncertainty: Measure confidence intervals, error rates, hallucination rates, bias, drift and failure modes. Define numerical production‑readiness criteria for each use case.
• Improve explainability: Use feature attribution, SHAP, LIME, error analysis, slice analysis and evaluation dashboards to explain model behavior and limitations.
• Translate results into business impact: Connect model metrics to revenue, cost, speed and risk. Quantify expected return, trade‑offs and downside scenarios for decision‑makers.

AI and Machine Learning Delivery

• Design complete AI solutions: Own problem framing, data strategy, modeling, evaluation, deployment and monitoring for machine learning, LLM and agentic systems.
• Build models for unstructured data: Develop production‑quality solutions using image, video, text, audio, sensor and time‑series data.
• Deliver computer vision solutions: Build detection, classification, segmentation, OCR and tracking systems with measurable performance benchmarks.
• Develop forecasting solutions: Create time‑series, demand and behavioral forecasting models with documented accuracy, error bands and integration into operational workflows.
• Build with foundation models: Use Claude and other LLMs for prompting, fine‑tuning, systematic evaluation and integration into multi‑step, tool‑using and multi‑agent workflows with memory and guardrails.

What We Are Looking For

• Bachelor's or Master's degree in Computer Science, Data Science, Machine Learning, Statistics or a related discipline, or equivalent practical experience.
• A track record of approximately 10 or more AI and machine learning projects deployed to production, with clear ownership of approach selection, evaluation and risk assessment.
• Strong statistical and algorithmic foundations, including hypothesis testing, experiment design, evaluation methodology and uncertainty quantification.
• Hands‑on experience with RAG, tool‑use patterns and agentic frameworks such as LangGraph, LlamaIndex, AutoGen or CrewAI, along with Model Context Protocol where relevant.
• Experience evaluating and protecting LLM and agentic systems through guardrails, systematic testing, hallucination measurement and failure analysis.
• Practical experience with explainability, bias and fairness assessment, model monitoring and drift detection.
• Familiarity with MLOps practices, including experiment tracking, CI/CD for machine learning, model registries and production monitoring.
• Experience with distributed training or inference optimization, including quantization, batching and GPU utilization.
• Working knowledge of Docker and Kubernetes.
• Strong communication skills, with the ability to present model results, risks and trade‑offs using clear numbers for engineers, business leaders and executives.

Technical Frameworks and Toolkit

• Deep Learning: PyTorch, TensorFlow, Keras, JAX and PyTorch Lightning.
• Generative AI and fine‑tuning: Hugging Face Transformers, PEFT, LoRA, QLoRA, TRL, Accelerate, DeepSpeed, bitsandbytes, Axolotl and Unsloth.
• Model serving: vLLM, TGI and Ollama.
• RAG and orchestration: LangChain and LlamaIndex.
• Agentic AI: Claude Agent SDK, Anthropic and OpenAI SDKs, LangGraph, AutoGen, CrewAI, Semantic Kernel, Model Context Protocol, tool calling and multi‑agent patterns.
• Computer vision: OpenCV, Detectron2, Segment Anything and image or video processing pipelines.
• Forecasting and optimization: statsmodels, Prophet, GluonTS, Darts, scikit‑learn, OR‑Tools, SciPy, PuLP, Gurobi and CVXPY.
• Evaluation and explainability: SHAP, LIME, LLM evaluation frameworks, error analysis, slice analysis and A/B testing.
• MLOps and infrastructure: MLflow, Weights & Biases, Docker, Kubernetes, CI/CD for machine learning, model registries and monitoring.

Compensation

Salary Range: US East/West Coast: $150000 - $170000

Disclaimer: The base salary range is a guideline and may vary based on factors such as candidate experience, specialized skills, and geographical location. Actual compensation may include additional benefits and bonuses.

Perks and Benefits

• Unlimited paid time off.
• Generous parental leave that exceeds typical industry standards.
• An entrepreneurial culture that supports thoughtful experimentation and calculated risk‑taking.
• Open communication with management and company leadership.
• Small, dynamic teams where individual contributions have visible impact.
• Medical, dental and vision coverage for employees.
• Access to disability and life insurance.
• Mental health and wellbeing support.
• Annual bonus program.
• Employer Stock Purchase Program.
• Annual team‑building experiences.
• Mentorship and sponsorship opportunities.
• Resources and support for people managers.

Cogniify is an equal opportunity employer. We celebrate diversity and are committed to creating an inclusive environment for all employees. All qualified applicants will receive consideration for employment without regard to race, color, religion, sex, sexual orientation, gender identity, national origin, disability, or any other protected characteristic