Senior Data Scientist

Position OverviewSenior-level data scientist role focused on building and deploying production NLP systems on bare metal infrastructure. This position requires a research-oriented mindset with the ability to build first-in-class products by translating cutting-edge research into innovative production solutions.Required QualificationsExperienceMinimum 5 years in data science/ML engineering rolesMinimum 3 years tenure in most recent organization in a relevant data science/ML roleProven track record of deploying ML models to productionExperience managing bare metal server infrastructureTechnical SkillsSQLAdvanced query optimization and performance tuningComplex joins, window functions, CTEsExperience with Snowflake, BigQuery, or RedshiftDatabase performance analysis and indexing strategiesNLP Technology StackTransformer architecturesRAG pipeline implementationLangChain, LlamaIndex, or similar frameworksVector databases: Pinecone, Weaviate, Chroma, FAISSModel fine-tuning: LoRA, QLoRAEmbedding models and semantic searchPrompt engineering techniquesProgramming & ML FrameworksPython (advanced level, production-grade code)PyTorch or TensorFlowHuggingFace Transformersscikit-learn, XGBoost, LightGBMKey ResponsibilitiesTechnical ExecutionDesign and implement production NLP solutions using state-of-the-art language modelsBuild and optimize complex SQL data pipelines processing millions of recordsDeploy ML models on bare metal GPU infrastructureConfigure and maintain GPU clusters for training and inferenceImplement MLOps practices: versioning, monitoring, automated retrainingOptimize model inference for latency and throughputTroubleshoot CUDA, driver, and hardware-level issuesSet up distributed training across physical serversResearch and prototype emerging ML techniquesLeadership & StrategyLead end-to-end ML projects from problem definition to production deploymentDrive innovation by researching and implementing first-in-class product featuresCoordinate cross-functional teams including data engineers, domain experts, and full-stack developers to deliver integrated solutionsDefine technical architecture and design decisions for ML systemsDrive adoption of ML best practices and engineering standards across teamsCollaborate with product and engineering leadership on ML roadmap and prioritiesPresent technical findings and recommendations to executive stakeholdersOwn critical ML infrastructure decisions and vendor evaluationsChampion innovation by evaluating and integrating cutting-edge ML researchLead cross-functional initiatives between data science, engineering, and product teamsFacilitate effective collaboration between technical and non-technical stakeholdersTranslate latest research papers into production-ready solutions.Required CompetenciesResearch-oriented mindset with ability to innovate and build first-in-class productsAbility to work independently with minimal supervision and drive projects autonomouslyStrong analytical and quantitative aptitudeExcellent problem-solving and logical reasoning skillsProven ability to collaborate with cross-functional teams (data engineers, domain experts, full-stack developers)Strong communication skills to translate technical concepts for non-technical stakeholdersWillingness to explore uncharted territory and experiment with novel approachesSelf-motivated with strong ownership mentalityStrong understanding of hardware constraints and optimizationAbility to work independently with bare metal infrastructureExperience with both cloud and on-premise deploymentsProven ability to take projects from research to productionTrack record of staying current with ML research and innovationsStrong debugging and troubleshooting skillsPreferred QualificationsExperience with pre-training multi-modal models (vision-language, audio-text, etc.)Hands-on experience with large-scale distributed training frameworks (DeepSpeed, FSDP, Megatron-LM)Contributions to open source ML projectsTechnical blog or active GitHub portfolioExperience with model quantization and efficient inferencePublications or conference presentationsKnowledge of multi-modal architectures (CLIP, Flamingo, GPT-4V style models)