Principal Forward Deployed GenAI Engineer / Architect

About Avanta LabsAvanta Labs helps small and mid-market companies deploy practical AI in weeks, not quarters. We build customer-owned systems spanning agentic AI, AI-native software, small language models, business transformation, strategy, governance, adoption, measurement, and advisory.About the Forward Deployed Engineering TeamWe combine senior technical judgment with hands-on delivery. Principals lead programs, develop engineers, and turn customer needs into reusable Avanta Labs platforms and IP.Role SummaryTranslate business objectives into AI programs, architect GenAI/ML systems, lead production delivery, and serve as the strategic technical escalation point.Role MissionBuild secure, scalable, economically defensible AI capabilities from executive objectives.What You Will DoArchitect AI systems across data, retrieval, models, agents, applications, evaluation, governance, and operations.Lead strategy through production, adoption, and managed operations.Select managed/open-weight models, traditional ML, fine-tuned/distilled models, or custom small language models based on evidence and economics.Evaluate prompt engineering, RAG, PEFT, pretraining, preference optimization, distillation, quantization, pruning, and custom training.Design domain-specific SLMs for privacy, edge, latency, and cost needs; guide data curation, labeling, lineage, and reproducibility.Architect training, inference, serving, and routing; optimize quality, latency, throughput, hardware use, reliability, and cost.Build safe single/multi-agent designs with tool contracts, state, memory, planning, permissions, human approvals, and containment of unsafe or unreliable behavior.Define offline/online, human, automated, adversarial, safety, and business evaluations, release gates, drift controls, rollback, and incident response.Lead security, identity, privacy, responsible AI, model risk, and supply-chain architecture.Evaluate build-versus-buy and vendors; establish reference architectures, platforms, and standards.Lead incidents, mentor engineers, and shape the product roadmap.Customer-Facing ResponsibilitiesLead discovery with executives, architects, security leaders, business owners, and delivery teams.Turn ambiguous goals into sequenced programs with measurable outcomes, architecture, governance, and operating models.Prioritize by value, feasibility, data readiness, risk, adoption, reuse, and cost.Present investment and architecture trade-offs; challenge weak, unsafe, or unjustified proposals.Negotiate scope, sequencing, constraints, customer responsibilities, and risk ownership.Guide adoption and production issues while maintaining trusted, hands-on customer relationships.Support strategic accounts, due diligence, proposals, and executive presentations.Technical ResponsibilitiesApply transformers, SFT, LoRA/QLoRA, adapters, preference optimization, distillation, quantization, pruning, and compression.Design distributed training/inference, checkpointing, lineage, experiment tracking, registries, and reproducible releases.Optimize serving with batching, caching, routing, cascades, speculative decoding, autoscaling, scheduling, and capacity planning.Design multitenant, event-driven, asynchronous, hybrid-cloud, edge, and disconnected systems.Establish CI/CD, MLOps, LLMOps, infrastructure/policy-as-code, resilience, recovery, and continuity.Implement IAM, authorization, data boundaries, sandboxing, isolation, audit, and policy controls for autonomous actions.Evaluate model, RAG, agent, safety, infrastructure, and business performance, including LLM-as-judge limitations; use A/B tests, drift monitoring, tracing, and root-cause analysis.Model full lifecycle cost; set acceptance criteria and ownership before go-live.What This Role Is Not Expected to OwnSales quotas or primarily sales engineering.Treating LLMs, agents, fine-tuning, or training as the answer to every problem.Detached advisory work without production involvement.Unilateral commercial, legal, or capital commitments beyond delegated authority.Required QualificationsMastery of Associate and Forward Deployed GenAI Engineer capabilities.Approximately 8–12+ years in engineering, ML, platforms, architecture, or technical leadership, including complex production AI/data systems.Hands-on code review, prototyping, full-stack diagnosis, and systems architecture.Leadership of major customer programs involving data, applications, security, operations, and change.Expertise in model selection, RAG, agents, evaluations, adaptation, performance, security, governance, and operations.Sound build-versus-buy, platform, infrastructure, and operating-model judgment under uncertainty.Credibility with boards, executives, and engineers; mentorship, crisis leadership, cross-team influence, and reusable IP creation.Equivalent experience may replace degree or tenure requirements.Preferred QualificationsFoundation-model or domain-specific SLM work; research, patents, open source, or advanced model engineering.GPU/accelerator operations, distributed training, or hardware-aware optimization.Regulated, safety-critical, privacy-sensitive, edge, or disconnected environments.Enterprise AI governance, internal AI platforms, or forward-deployed practice design.Confidential computing, federated learning, or privacy-preserving ML.Vendor, partnership, investment, or acquisition evaluation; board/industry presentations.Representative TechnologiesEquivalent technologies welcome; not all products are required.Languages: Python; TypeScript/JavaScript; Java; C#; Go; SQL.Cloud/models: AWS Bedrock/SageMaker; Azure AI Foundry/OpenAI; Vertex AI; OpenAI; Anthropic; Hugging Face; open-weight models.Model engineering: PyTorch; TensorFlow; JAX; Transformers; PEFT; DeepSpeed; Accelerate; Ray.Inference: vLLM; TGI; NVIDIA Triton; ONNX; TensorRT; MLX; gateways and serving.Agents: LangGraph; Semantic Kernel; AutoGen; CrewAI; LlamaIndex; SDKs; tool protocols; sandboxing.Data/retrieval: PostgreSQL/pgvector; OpenSearch/Elasticsearch; Pinecone; Weaviate; Milvus; Qdrant; Redis; Neo4j; streaming/lakehouse/document systems.Evaluation: MLflow; Weights & Biases; LangSmith; Arize Phoenix; OpenTelemetry; Prometheus; Grafana.Infrastructure/security: Docker; Kubernetes; Terraform; CI/CD; Argo CD; serverless; service meshes; IAM; RBAC/ABAC; encryption; DLP; audit; guardrails.Core Competencies and Soft SkillsArchitectural judgment, executive communication, commercial awareness.Customer empathy and influence under uncertainty.Mentorship, crisis leadership, negotiation, and stakeholder management.Clear writing, risk judgment, and hands-on leadership.Converting delivery knowledge into reusable standards and IP.Scope, Autonomy, and Decision-MakingLead architecture for assigned strategic engagements and approve standards, reference patterns, release gates, and platform choices within CTO-delegated authority. Escalate material company investment, legal, customer-risk, or product-strategy decisions to Avanta Labs' CTO and executive team.How Success Will Be MeasuredDeliver production AI systems linked to measurable customer outcomes.Make sound model, architecture, platform, and security decisions.Reduce delivery time, operational risk, and complexity using reusable standards.Establish evaluation, governance, observability, and release practices.Develop technical leaders and resolve severe drift, quality, latency, cost, security, or reliability issues.Influence customer strategy, create reusable IP, and build executive trust.Expected Outcomes During the First 6–12 MonthsLead a strategic program from executive framing through production.Improve a reference architecture, evaluation standard, governance control, or delivery platform.Standardize build-versus-buy and model-selection decisions.Develop an engineer to independently lead a complex workstream.Turn repeated customer needs into a reusable accelerator or product-roadmap proposal.Demonstrate gains in quality, risk, customer value, cost, or deployment speed.Represent Avanta Labs in executive and architecture reviews.Travel or Customer-Site ExpectationsTravel may be required for discovery, workshops, executive reviews, delivery, and escalations; final expectations will be confirmed before hire.Reporting RelationshipReports to an Avanta Labs technical leader designated by executive leadership. May lead engineering pods and mentor engineers across engagements.Compensation and BenefitsCompensation and benefits are to be confirmed. Avanta Labs will share employment structure, compensation, healthcare, paid time off, and benefits before an offer.Equal Opportunity and Accessibility StatementAvanta Labs is an equal opportunity employer. We consider qualified applicants without regard to any legally protected status and provide reasonable accommodations during recruiting and employment. Candidates may request accommodations when applying or interviewing.