Job Specifications
In this role, you will contribute to building and shipping AI features end-to-end: data prep, modeling, evaluation, deployment, and iteration. Additionally, collaborate with the engineering team to translate ideas and research into reliable, production-grade systems.
What You’ll Do
Build LLM-powered features (prompt design, RAG pipelines, tools/plug-ins, evaluations, guardrails).
Experiment with agentic AI patterns (tool use, planning/re-planning, multi-agent workflows) and ship reliable agents.
Implement and evaluate machine-learning models (classification, regression, clustering, NLP, CV) from prototype to production.
Write clean, well-tested Python code for data processing, modeling, and service APIs.
Package and deploy models/services on AWS (e.g., S3, Lambda, ECS/EKS, SageMaker) with basic CI/CD.
Design simple, efficient data pipelines and integrate with databases (SQL/NoSQL) and vector stores.
Monitor models in production (latency, drift, quality) and iterate based on telemetry and user feedback.
Read papers/blogs/specs and quickly translate ideas into working prototypes.
What You’ll Bring
Strong foundation in algorithms and data structures; able to analyze time/space complexity and choose the right approach.
Solid understanding of core ML principles: bias/variance, feature engineering, cross-validation, regularization, evaluation metrics.
Familiarity with LLMs: tokenization basics, model families, fine-tuning concepts, RAG patterns, and LLM evaluations.
Exposure to agentic AI concepts: tool calling, planning, memory, and simple multi-agent orchestration.
Knowledge of Model Context Protocol (MCP) for context sharing, secure integrations, and tool orchestration.
Proficiency in Python and common libraries (NumPy, pandas, scikit-learn; plus, PyTorch or TensorFlow preferred).
Comfort with AWS fundamentals (IAM, S3, compute/container runtimes) or equivalent cloud experience.
Experience with databases: writing efficient SQL, understanding normalization/indices; basic NoSQL (e.g., DynamoDB) awareness.
Familiarity with vector databases (e.g., FAISS, Pinecone, Milvus) is a plus.
Version control (Git) and basic software craftsmanship (testing, linting, code reviews).
Nice to have
Data engineering basics: Airflow/Prefect, message queues, data validation.
API development (FastAPI/Flask) and simple observability (logs/metrics/traces).
Security, privacy, and responsible-AI awareness (PII handling, prompt injection basics, red-teaming mindset).
Math comfort: linear algebra, probability, calculus.
How You Work
Strong ability to adapt to new technologies and rapidly learn by reading docs/papers and implementing new ideas.
Bias for action: iterate quickly, measure results, and improve based on evidence.
Clear communication and collaborative mindset; comfortable receiving and giving feedback.
Qualifications
Bachelor’s or master's in computer science, Data Science, EE, or related field (or equivalent projects/internships).
0–2 years of professional experience; internships, open-source, or notable personal projects count.
Smith is an equal opportunity employer
VEVIRAA Federal Contractor
We are an Equal Opportunity/Affirmative Action Employer.