Description :

Build and Own the AI Architecture :

- Design and maintain model pipelines (LLMs, embeddings, RAG, vector DB, orchestrators).

- Create robust retrieval systems over internal data (Docs, Notion, Slack, Google Drive, Dropbox, and email extracts).

Develop Production-Grade AI Agents :

- Architect multi-agent systems using frameworks like LangChain, LlamaIndex, or custom orchestration.

- Build autonomous or semi-autonomous agents for sourcing, diligence, research, and internal ops.

- Deploy tools with monitoring, evaluations, and feedback loops.

Lead the Internal AI Product Roadmap :

- Work with partners, analysts, and the platform team to identify opportunities for automation or augmentation.

- Translate business processes into end-to-end AI products.

Own Quality, Testing, and Evaluation :

- Build evaluation suites for accuracy, relevance, hallucinations, and latency.

- Implement offline and online eval frameworks for prompts, pipelines, and RAG systems.

Manage Deployments and Infrastructure :

- Set up CI/CD, observability, logging, and agent monitoring.

- Experimentation and Model Optimization :

- Run fast iterations on prompting, fine-tuning, LoRA, custom embeddings, or domain-specific models.

- Evaluate and integrate new LLMs, vector DBs, and inference optimizations.

Cross-Functional Leadership :

- Partner with the investment team to productize workflows.

- Collaborate with data engineering to ensure data quality and pipelines.

- Work with leadership to make buy/build/partner decisions.

Long-Term Architecture and Governance :

- Define standards for data security, privacy, safety, and human-in-the-loop decisioning.

- Own documentation, design reviews, and technical guidance.

Requirements :

- 5-8+ years in engineering, with 1-2 years in GenAI.

- Strong with LLMs, RAG, vector DBs, and embeddings.

- Experience deploying AI models to production.

- Hands-on experience with PyTorch/JAX and GPUs is a plus.

- Prefer candidates from top AI startups/labs.

Nice to Have :

- Experience fine-tuning or distilling LLMs (LoRA, SFT).

- Background in startups, VC, SaaS, or high-velocity environments.