Full Stack Developer - AI platform
Posted 1 week ago by Twine
Negotiable
Undetermined
Remote
United Kingdom
Summary: The role of a freelance full stack developer involves establishing a production-grade Pinecone vector database and backend for knowledge retrieval within an AI-powered maritime intelligence platform. The developer will design and implement a robust retrieval stack that integrates maritime data with PostgreSQL, ensuring low-latency search capabilities and secure APIs. The position requires a blend of data engineering skills and product-minded execution, with a focus on documentation and maintainability. Work is fully remote, with collaborative scope and milestone definitions.
Key Responsibilities:
- Design and provision a Pinecone index with appropriate dimensionality, metric, and replication/sharding strategy.
- Create metadata schema and namespaces for faceted filtering and multi-tenant separation.
- Develop an embedding pipeline including chunking strategy, text normalization, deduplication, and embedding generation.
- Implement ingestion & sync service to upsert vectors from PostgreSQL and external maritime sources.
- Build a minimal reference app demonstrating query, filter, and retrieval flows.
Key Skills:
- Experience building vector-based retrieval systems using Pinecone.
- Proficiency with PostgreSQL and integrating it as the source of truth for embedding pipelines.
- Familiarity with LLM/RAG patterns and embedding model selection.
- Cloud fluency (AWS/GCP/Azure) and containerization (Docker).
- Data engineering skills including streaming/batch ingestion and job orchestration.
- Knowledge of security best practices for secrets and access control.
- Excellent communication and ability to work asynchronously in a remote setup.
Salary (Rate): undetermined
City: undetermined
Country: United Kingdom
Working Arrangements: remote
IR35 Status: undetermined
Seniority Level: undetermined
Industry: IT
An AI-powered maritime intelligence platform requires a freelance full stack developer to establish a production-grade Pinecone vector database and supporting backend for knowledge retrieval. The role focuses on designing and implementing a robust retrieval stack (RAG) that ingests maritime data, stores high-quality embeddings with rich metadata, and exposes secure, scalable APIs for downstream applications. The solution should integrate with PostgreSQL as a system of record, support efficient updates, and deliver low-latency semantic and hybrid search. The ideal candidate combines data engineering pragmatism with product-minded execution, ensuring documentation, observability, and maintainability are first-class. Work is fully remote, with scope and milestones defined collaboratively.
Deliverables
- Design and provision a Pinecone index (pods/collections) with appropriate dimensionality, metric, and replication/sharding strategy for target latency and scale.
- Metadata schema and namespaces enabling faceted filtering (e.g., vessel type, geography, timestamp, source) and multi-tenant separation where needed.
- Embedding pipeline: chunking strategy, text normalization, deduplication, PII handling, and embedding generation (OpenAI/Cohere/Voyage or equivalent), with retries and backoff.
- Ingestion & sync service to upsert vectors from PostgreSQL and external maritime sources (e.g., AIS feeds, vessel registries, reports), including change-data-capture or periodic batch jobs.
- Minimal reference app (e.g., Next.js/Node or Python/FastAPI) demonstrating query, filter, and retrieval flows, including API keys and role-based access.
Requirements
- Demonstrable experience building vector-based retrieval systems using Pinecone (index design, metadata filtering, upserts, query optimization).
- Strong proficiency with PostgreSQL (schema design, indexing, ETL/ELT) and integrating it as the source of truth for embedding pipelines.
- Familiarity with LLM/RAG patterns: embedding model selection, chunking heuristics, reranking, caching, and evaluation of retrieval quality.
- Cloud fluency (AWS/GCP/Azure), containerization (Docker), CI/CD, and infrastructure-as-code.
- Data engineering skills: streaming/batch ingestion, file formats (JSONL/Parquet), and job orchestration (e.g., Airflow, Temporal, or serverless schedulers).
- Security best practices for secrets, token management, and access control; comfort with monitoring/observability tooling.
- Excellent communication, proactive status updates, and ability to work asynchronously in a remote setup.
About Twine
Twine is a leading freelance marketplace connecting top freelancers, consultants, and contractors with companies needing creative and tech expertise. Trusted by Fortune 500 companies and innovative startups alike, Twine enables companies to scale their teams globally.
Our Mission
Twine's mission is to empower creators and businesses to thrive in an AI-driven, freelance-first world.