Senior GenAI Full Stack Engineer (Python + Typescript)

Location: Remote-First

Level: Senior (5+ years) | Full-time

About DevSavant

DevSavant is the engineering powerhouse behind Savant Growth's portfolio companies. We're not your typical dev shop—we pair small, surgical teams with cutting-edge GenAI tooling to solve real business problems. Our unique approach: ship experimental features in focused one-week innovation sprints, then mature the winners into robust production services that scale.

The Opportunity

We're seeking a senior, self-directed GenAI Full-Stack Engineer who thrives at the intersection of rapid innovation and production excellence. You'll own the complete lifecycle of GenAI initiatives:

Prototype Fast – Design, build, and demo proof-of-concepts within single sprints

Pressure-Test & Iterate – Instrument, evaluate, and refine until ideas prove value

Productionize – Apply MLOps, observability, and CI/CD so successful PoCs scale

The ideal teammate moves quickly with minimal supervision, selects the right tool for each job, and communicates clearly across engineering and product stakeholders.

What You'll Build

Core Responsibilities

• Architect and develop LLM-powered workflows using LangGraph, LangChain, and emerging frameworks

• Design and implement production RAG systems with hybrid search, re-ranking, and semantic caching

• Build multi-agent orchestration with CrewAI, AutoGen, AgentKit—selecting build-vs-buy per use case

• Create real-time AI interfaces with streaming responses, progressive enhancement, and edge deployment

• Connect GenAI pipelines to enterprise data via low-code orchestrators (n8n, Make, Langflow)

• Implement formal prompt lifecycles with version control, A/B testing, and performance tracking

• Add runtime guardrails and automated evaluation using Guardrails-AI, Ragas, Promptfoo

• Optimize for cost & latency via prompt caching, selective chunking, and model routing

• Instrument with enterprise observability (LangSmith, OpenTelemetry, Prometheus/Grafana)

Production Excellence

When prototypes show traction, you'll:

• Design scalable architectures handling 100K+ concurrent users

• Implement comprehensive testing (unit, integration, E2E, evaluation suites)

• Build CI/CD pipelines with automated quality gates and progressive rollouts

• Deploy with container orchestration (Docker, Kubernetes) and auto-scaling

• Monitor AI-specific metrics (latency, cost per request, quality scores)

Technical Requirements

Languages & Frameworks

• Python (5+ years): Advanced proficiency with type hints, async/await, dataclasses

• TypeScript/JavaScript (5+ years): Modern ES2022+, React 18+, Node.js 20+

• Full-Stack Frameworks: Next.js 14+ (App Router), Remix, FastAPI, Express.js

GenAI & ML Stack

• LLM Orchestration: LangGraph, LangChain, LlamaIndex (2+ years production)

• Multi-Agent Systems: CrewAI, AutoGen, Autogen Studio, custom agent frameworks

• Vector Databases: Pinecone, Weaviate, Qdrant, pgvector (with hybrid search)

• RAG Expertise: Advanced patterns (HyDE, multi-hop, contextual compression)

• Model Integration: OpenAI, Anthropic, Mistral, Llama, multimodal models

• Evaluation & Testing: Ragas, Promptfoo, custom eval frameworks

Infrastructure & Operations

• Cloud Platforms: AWS/GCP/Azure with AI services (Bedrock, Vertex AI, Azure OpenAI)

• Containerization: Docker, Kubernetes, serverless architectures

• Observability: LangSmith, DataDog, OpenTelemetry distributed tracing

• MLOps: Model versioning, A/B testing, feature stores, experiment tracking

• Low-Code Tools: n8n, Make, Zapier for rapid integration

Data & Performance

• Databases: PostgreSQL, Redis, DynamoDB, vector stores

• Streaming: Kafka, Redis Streams, WebSockets for real-time AI

• Caching: Multi-tier caching strategies, CDN optimization

• Security: OAuth2, JWT, API gateways, prompt injection prevention

Experience & Mindset

Required Experience

• 5+ years professional full-stack development

• 2+ years shipping GenAI/LLM features to production

• 1+ year with vector databases and production RAG systems

• Proven track record of 0→1 product development

• Experience with high-traffic applications (10K+ concurrent users)

Working Style

• Autonomous & Accountable – Set your course, surface blockers early, deliver independently

• Bias for Action – Spike new libraries in the morning, demo results by afternoon

• Quality Owner – Comprehensive testing mindset with evaluation-driven development

• Continuous Learner – Track bleeding-edge releases, introduce relevant tools

• Strong Communicator – Translate complex AI concepts to all stakeholders

P

referred Qualifications

• Experience with fine-tuning (LoRA/QLoRA) on Mistral, Llama, or domain-specific models

• Multimodal AI applications (vision + language, audio processing)

• Research implementation – turning papers into production code

• Open source contributions to major AI/ML projects

• Published technical content (blogs, talks, papers) on GenAI topics

• Experience with AI safety and alignment practices