AI/ML Solution Architect
AceStack LLC, Raritan, NJ, United States
Job Title : AI/ML Solution Architect
Location : Raritan, NJ (ONSITE)
FULLTIME ONLY
Job Description
Must Have Technical/Functional Skill
Python (PyTorch/TensorFlow), SQL, API design; familiarity with LangChain/LlamaIndex, Azure/AWS AI, GitHub, Docker, Kubernetes
Roles & Responsibilities
• Define end-to-end AI/ML and GenAI architectures (data ingestion → feature store → model training → inference → monitoring).
•
Design LLM-based solutions:
RAG pipelines, prompt orchestration, guardrails, policy enforcement, and human-in-the-loop workflows.
• Select appropriate models (open source vs. hosted foundation models), vector databases, embedding strategies, and retrieval frameworks.
•
Author architecture artifacts:
context diagrams, sequence diagrams, deployment topology, decision records, security controls, and data lineage.
• Lead data architecture for AI workload data quality, labeling, metadata, PII handling, and de-identification.
•
Establish MLOps:
CI/CD for models, feature store, model registry, automated evaluation, canary rollout, A/B testing, drift monitoring.
• Optimize cost and performance for training/inference (GPU/TPU selection, quantization, distillation, batching).
• Ensure data privacy, consent, and secure access (role-based control, encryption, KMS, secret management).
Location : Raritan, NJ (ONSITE)
FULLTIME ONLY
Job Description
Must Have Technical/Functional Skill
Python (PyTorch/TensorFlow), SQL, API design; familiarity with LangChain/LlamaIndex, Azure/AWS AI, GitHub, Docker, Kubernetes
Roles & Responsibilities
• Define end-to-end AI/ML and GenAI architectures (data ingestion → feature store → model training → inference → monitoring).
•
Design LLM-based solutions:
RAG pipelines, prompt orchestration, guardrails, policy enforcement, and human-in-the-loop workflows.
• Select appropriate models (open source vs. hosted foundation models), vector databases, embedding strategies, and retrieval frameworks.
•
Author architecture artifacts:
context diagrams, sequence diagrams, deployment topology, decision records, security controls, and data lineage.
• Lead data architecture for AI workload data quality, labeling, metadata, PII handling, and de-identification.
•
Establish MLOps:
CI/CD for models, feature store, model registry, automated evaluation, canary rollout, A/B testing, drift monitoring.
• Optimize cost and performance for training/inference (GPU/TPU selection, quantization, distillation, batching).
• Ensure data privacy, consent, and secure access (role-based control, encryption, KMS, secret management).