About Role :

We're looking for an Applied AI Engineer who operates at the intersection of LLMs, ML Engineering, and Data Engineering - someone who can build and ship production-grade AI systems, not just prototypes.

This is a high-impact opportunity to own the foundation model layer, build end-to-end model pipelines, apply classical ML for forecasting/anomaly detection, and architect the data infrastructure powering it all.

What You'll Work On :

Foundation Models & Fine-Tuning

1. Fine-tune open-weight LLMs (Llama, Qwen, Mistral, Gemma) using LoRA/QLoRA, DPO/ORPO, continued pre-training

2. Build evaluation harnesses, regression suites, and reliable inference systems

3. Own serving stack using vLLM / TGI / SGLang on GKE or Vertex AI

4. Improve reliability, latency, structured outputs, and model performance in production

LLM Pipelines & Agent Systems :

1. Design pipelines across training - evaluation - deployment - monitoring

2. Build RAG and retrieval systems over large domain-specific corpora

3. Instrument drift detection, prompt/response monitoring, latency SLOs, and cost attribution

Classical ML & Forecasting :

1. Build models for liquidity forecasting, partner behavior modeling, and anomaly detection

2. Apply the right modeling approach - XGBoost, time-series, or fine-tuned transformers - based on the problem

Data Engineering for ML :

1. Build feature store infrastructure with online/offline parity and point-in-time correctness

2. Create scalable ETL pipelines and model-ready datasets

3. Own schema design, data quality, PII handling, synthetic data, and feature monitoring

What We're Looking For :

1. 6-8 years building ML systems in production

2. Strong hands-on expertise with transformer models and LLM fine-tuning

3. Experience shipping fine-tuned LLMs to production

4. Strong Python + PyTorch + Hugging Face ecosystem

5. Experience with inference serving (vLLM / TGI / SGLang)

6. Solid foundation in classical ML, forecasting, and statistical reasoning

7. Strong data engineering skills (SQL, orchestration, schema design)

8. Experience with GCP (or equivalent AWS/Azure stack)

Strong Plus :

1. Agent frameworks / tool-use evaluation / multi-agent systems

2. Distributed training (FSDP, DeepSpeed)

3. Fintech, payments, fraud or regulated domain experience

4. Open-source ML/LLM contributions

5. Liquidity / treasury forecasting exposure

You'll Thrive Here If :

1. You think in systems, not just notebooks

2. You care about production reliability, not demoware

3. You can build from scratch and operate in an early-stage environment

4. You like solving hard problems where model decisions have real business impact