AlignBits LLC

Company Description At AlignBits LLC, we turn the digital dreams of our clients into reality. We work closely with our users throughout the development process to ensure that our goals remain aligned. We are committed to producing exceptional software for each client. Role Description This is a full-time remote role for a Generative AI Engineer. The Generative AI Engineer will be responsible for designing and developing AI models, conducting research to improve the performance of generative models, and collaborating with cross-functional teams. Day-to-day tasks include implementing machine learning algorithms, testing model performance, and staying updated with the latest advancements in AI technology. Tech Stack (you don’t need all, but strong coverage is expected) LLM/RAG: Python, LangChain/LlamaIndex (or equivalent), OpenAI/Anthropic/HF, embeddings, VectorDBs (Pinecone, Weaviate, Milvus, pgvector). Modeling: Fine-tuning, instruction tuning, RLHF/DPO (nice-to-have), evaluation frameworks. Data/Parsing: pdfplumber, PyMuPDF, LayoutLM/Donut/DocTR (or similar), regex/grammar parsers. Agents & Scraping: Playwright/Puppeteer/Selenium, asyncio, proxy management, scraping ethics/robots compliance. Backend/MLOps: FastAPI/Flask, Docker, AWS (EC2/ECS/S3/RDS/Lambda), CI/CD, monitoring (Prometheus/Grafana), feature flags, experiment tracking. Security/Privacy: PII handling, data anonymization, access controls, key management. Qualifications 2+ years in ML/AI software engineering (at least 1+ years hands-on with LLMs/RAG). Proven delivery of an LLM or search/retrieval system to production (not just prototypes). Strong Python engineering, testing discipline, and systems thinking. Experience designing evaluation harnesses and closing the loop with retraining. Excellent communication; able to work independently as the founding AI hire and mentor others. Nice to Have Experience in supply chain/logistics or B2B data integrations. Schema mapping, ETL, or EDI/X12/EDIFACT exposure. Document AI (layout understanding, OCR) and schema-aware validation. Cost/performance optimization of inference workloads.