Vice President AI Fixed Income Quantitive Developer
Here's the plain text spec:
AI Research Engineer — Fixed Income Hedge Fund | New York, NY | On-Site
ABOUT THE FIRM
Our client is a fundamentals-driven fixed income hedge fund managing institutional capital across rates, credit, and structured products. The investment process is analytically rigorous — not a pod shop running factor tilts. They are making a deliberate, well-resourced push into applied AI to sharpen their edge across research, signal generation, and portfolio analytics. The person they hire will be the tip of that spear.
WHAT YOU'LL OWN
Design and deploy LLM-powered tools supporting fixed income research workflows — credit document parsing, earnings call analysis, macro data synthesis, issuer monitoring. Build production-grade pipelines: RAG, fine-tuning workflows, prompt engineering frameworks, and evaluation harnesses — not demos, not notebooks. Work directly with PMs and analysts to translate investment intuition into system requirements. Own infrastructure decisions around model serving, inference optimization, and hardware allocation — GPU vs. ASIC tradeoffs, quantization strategies, batching, latency/cost optimization. Stay current on model capabilities, benchmark meaningfully against investment use cases, and make deployment recommendations with a practitioner's discipline.
BACKGROUND
3–4+ years building and shipping ML/AI systems in production — buyside or fintech strongly preferred. Strong academic grounding in ML or NLP; a graduate degree in CS or a related quantitative field is a meaningful plus. Genuine fluency in LLMs: fine-tuning, RLHF, RAG architectures, context management — you've built with these, not just read about them. Hardware literacy that goes beyond buzzwords — GPU memory hierarchies, CUDA fundamentals, and where custom silicon (ASICs, TPUs) makes economic and performance sense vs. commodity GPU clusters. Fixed income fluency across rates, credit, and structured products — you don't need to be a PM, but you need to speak the language. Python-native; comfortable with inference frameworks (vLLM, TGI, TensorRT-LLM), cloud ML infrastructure, and the orchestration layer. Prior experience building internal tools for traders or analysts is a strong differentiator.
THE PERSON
A builder who knows markets, or a markets person who has gone deep enough into ML infrastructure to ship real systems. Either path works. What doesn't work: an academic who wants to run experiments without shipping, or an engineer who needs the investment context handed to them. The fund is small enough that your fingerprints will be all over what gets built. The team is serious, the capital base is institutional, and they're committed to this initiative — not running a one-year pilot with a sunset clause.